From ranjeet.rai at orkash.com Wed Apr 1 00:49:50 2009 From: ranjeet.rai at orkash.com (ranjeet.rai at orkash.com) Date: Wed, 01 Apr 2009 13:19:50 +0530 Subject: [cAos] IP bonding in CAOS Message-ID: <49D31C9E.1010404@orkash.com> Hi, I am new to CAOS NSA, looking forward to configure IP bonding. I have configured all the settings and have created the bond0 file with proper entries. This same setting is working fine on CentOS. When I restart the network services get the error message : " I do not understand the TYPE "Ethernet" Please, help me to solve this. Regards, Ranjeet Rai From gmkurtzer at gmail.com Wed Apr 1 09:28:14 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Wed, 1 Apr 2009 09:28:14 -0700 Subject: [cAos] IP bonding in CAOS In-Reply-To: <49D31C9E.1010404@orkash.com> References: <49D31C9E.1010404@orkash.com> Message-ID: <571f1a060904010928u39e6d0e9k981b7af56b0572b3@mail.gmail.com> To enable network device bonding, you need to create a file in /etc/sysconfig/nics/ with the name of the configuration you wish to use (e.g. bond0). Then add the following information to the file called bond0: DEVICE=bond0 ONBOOT=yes TYPE=bond ENSLAVE=eth0,eth1,eth2 BOOTPROTO=static IPADDR=10.0.0.1 NETMASK=255.255.255.0 Obviously, change the necessary parts. Then run the command: # /sbin/ifup bond0 Let us know how this works for you and if you had to take any additional steps to get it working. Greg On Wed, Apr 1, 2009 at 12:49 AM, ranjeet.rai at orkash.com wrote: > Hi, > > I am new to CAOS NSA, looking forward to configure IP bonding. I have > configured all the settings and have created the bond0 file with proper > entries. This same setting is working fine on CentOS. > > When I restart the network services get the error message : " I do not > understand the TYPE "Ethernet" > > Please, help me to solve this. > > Regards, > > Ranjeet Rai > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From glykos at mbg.duth.gr Thu Apr 2 03:15:44 2009 From: glykos at mbg.duth.gr (Nicholas M Glykos) Date: Thu, 2 Apr 2009 13:15:44 +0300 (EEST) Subject: [cAos] IP bonding in CAOS In-Reply-To: <571f1a060904010928u39e6d0e9k981b7af56b0572b3@mail.gmail.com> References: <49D31C9E.1010404@orkash.com> <571f1a060904010928u39e6d0e9k981b7af56b0572b3@mail.gmail.com> Message-ID: Dear All, I do not mean to hijack the thread, but is there a consensus in the community with respect to channel bonding vs. keeping separate managerial+data interfaces ? I understand that this depends on the application(s) running on the cluster, but I was just wondering what is the most commonly used approach. Thanks, Nicholas -- Dr Nicholas M. Glykos, Department of Molecular Biology and Genetics, Democritus University of Thrace, University Campus, Dragana, 68100 Alexandroupolis, Greece, Tel/Fax (office) +302551030620, Ext.77620, Tel (lab) +302551030615, http://utopia.duth.gr/~glykos/ From gmkurtzer at gmail.com Thu Apr 2 12:39:11 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Thu, 02 Apr 2009 12:39:11 -0700 Subject: [cAos] IP bonding in CAOS In-Reply-To: References: <49D31C9E.1010404@orkash.com> <571f1a060904010928u39e6d0e9k981b7af56b0572b3@mail.gmail.com> Message-ID: <49D5145F.3060300@gmail.com> Another thing to consider is MPI support for multi-rail. Personally, we have implemented all three depending on the specifics about the application, architecture requirements and budget. Sorry not to give a straight answer as it can be an apples Vs. oranges comparison. Greg Nicholas M Glykos wrote: > Dear All, > > I do not mean to hijack the thread, but is there a consensus in the > community with respect to channel bonding vs. keeping separate > managerial+data interfaces ? I understand that this depends on the > application(s) running on the cluster, but I was just wondering what is > the most commonly used approach. > > Thanks, > Nicholas > > > From gmkurtzer at gmail.com Thu Apr 2 12:41:32 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Thu, 02 Apr 2009 12:41:32 -0700 Subject: [cAos] mdadm etc. in caos In-Reply-To: <20090326160716.GL22686@loop.mdy.univie.ac.at> References: <20090326160716.GL22686@loop.mdy.univie.ac.at> Message-ID: <49D514EC.6090306@gmail.com> Just wanted to let you know that we didn't miss this email. Thanks for your patience. Greg Stefan Boresch wrote: > [Although the problem shows on a machine which is provisioned by perceus, > I believe this is mostly a caos problem!] > > I want to boot a machine with caos that aside from the system disk has > a fairly large software raid0 created under Ubuntu with mdadm, which I don't > want to nuke. Using the old mdadm.conf file (created by Ubuntu's tools), > I can assemble the raid manually with > > mdadm -As /dev/md0 > > However, I fail on various levels with the caos provided > /etc/init.d/mdmonitor and /etc/init.d/mdmpd scripts. First, the > daemon() function in /etc/init.d/functions doesn't seem to > understand options, i.e., the line > > daemon --check --user=root mdadm ${OPTIONS} > > from /etc/init.d/mdmonitor results in > > /etc/init.d/mdmonitor start > Starting --check:pidof: invalid options on command line! > > error > /etc/rc.d/init.d/functions: line 90: --check: command not found > > Second, removing --check --user=root, gives me an OK, but no usable > raid. The script checks whether there is info in /etc/mdadm.conf, but > doesn't seem to use it. I would guess the first problem is the result > of a mismatch between redhat scripts and the functions available on > caos, but what am I missing in the second step? > > Thanks in advance, > > Stefan > > From ranjeet.rai at orkash.com Thu Apr 2 23:00:12 2009 From: ranjeet.rai at orkash.com (ranjeet.rai at orkash.com) Date: Fri, 03 Apr 2009 11:30:12 +0530 Subject: [cAos] cAos Digest, Vol 24, Issue 2 In-Reply-To: References: Message-ID: <49D5A5EC.5040403@orkash.com> Hi Greg, This is working fine now, I just changed the TYPE=bond. Thanks for your great help. Regards, Ranjeet caos-request at caoslinux.org wrote: > Send cAos mailing list submissions to > caos at caoslinux.org > > To subscribe or unsubscribe via the World Wide Web, visit > http://lists.caosity.org/mailman/listinfo/caos > or, via email, send a message with subject or body 'help' to > caos-request at caoslinux.org > > You can reach the person managing the list at > caos-owner at caoslinux.org > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of cAos digest..." > > > Today's Topics: > > 1. Re: IP bonding in CAOS (Greg Kurtzer) > 2. Re: IP bonding in CAOS (Nicholas M Glykos) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Wed, 1 Apr 2009 09:28:14 -0700 > From: Greg Kurtzer > Subject: Re: [cAos] IP bonding in CAOS > To: Community-Assembled Operating System > Message-ID: > <571f1a060904010928u39e6d0e9k981b7af56b0572b3 at mail.gmail.com> > Content-Type: text/plain; charset=ISO-8859-1 > > To enable network device bonding, you need to create a file in > /etc/sysconfig/nics/ with the name of the configuration you wish to > use (e.g. bond0). Then add the following information to the file > called bond0: > > DEVICE=bond0 > ONBOOT=yes > TYPE=bond > ENSLAVE=eth0,eth1,eth2 > BOOTPROTO=static > IPADDR=10.0.0.1 > NETMASK=255.255.255.0 > > Obviously, change the necessary parts. > > Then run the command: > > # /sbin/ifup bond0 > > Let us know how this works for you and if you had to take any > additional steps to get it working. > > Greg > > > > On Wed, Apr 1, 2009 at 12:49 AM, ranjeet.rai at orkash.com > wrote: > >> Hi, >> >> I am new to CAOS NSA, looking forward to configure IP bonding. I have >> configured all the settings and have created the bond0 file with proper >> entries. This same setting is working fine on CentOS. >> >> When I restart the network services get the error message : " I do not >> understand the TYPE "Ethernet" >> >> Please, help me to solve this. >> >> Regards, >> >> Ranjeet Rai >> _______________________________________________ >> cAos mailing list >> cAos at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/caos >> >> > > > > From gmkurtzer at gmail.com Thu Apr 2 23:02:30 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Thu, 2 Apr 2009 23:02:30 -0700 Subject: [cAos] cAos Digest, Vol 24, Issue 2 In-Reply-To: <49D5A5EC.5040403@orkash.com> References: <49D5A5EC.5040403@orkash.com> Message-ID: <571f1a060904022302t65e5b0c0mb4aee7ad53341d19@mail.gmail.com> Glad to help! I also added this page to the wiki: http://wiki.caoslinux.org/Network_Configuration Everyone should feel free to check it out and make corrections if necessary. Thanks, Greg On Thu, Apr 2, 2009 at 11:00 PM, ranjeet.rai at orkash.com wrote: > Hi Greg, > > This is working fine now, I just changed the TYPE=bond. > > Thanks for your great help. > > Regards, > > Ranjeet > > caos-request at caoslinux.org wrote: >> Send cAos mailing list submissions to >> ? ? ? caos at caoslinux.org >> >> To subscribe or unsubscribe via the World Wide Web, visit >> ? ? ? http://lists.caosity.org/mailman/listinfo/caos >> or, via email, send a message with subject or body 'help' to >> ? ? ? caos-request at caoslinux.org >> >> You can reach the person managing the list at >> ? ? ? caos-owner at caoslinux.org >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of cAos digest..." >> >> >> Today's Topics: >> >> ? ?1. Re: IP bonding in CAOS (Greg Kurtzer) >> ? ?2. Re: IP bonding in CAOS (Nicholas M Glykos) >> >> >> ---------------------------------------------------------------------- >> >> Message: 1 >> Date: Wed, 1 Apr 2009 09:28:14 -0700 >> From: Greg Kurtzer >> Subject: Re: [cAos] IP bonding in CAOS >> To: Community-Assembled Operating System >> Message-ID: >> ? ? ? <571f1a060904010928u39e6d0e9k981b7af56b0572b3 at mail.gmail.com> >> Content-Type: text/plain; charset=ISO-8859-1 >> >> To enable network device bonding, you need to create a file in >> /etc/sysconfig/nics/ with the name of the configuration you wish to >> use (e.g. bond0). Then add the following information to the file >> called bond0: >> >> DEVICE=bond0 >> ONBOOT=yes >> TYPE=bond >> ENSLAVE=eth0,eth1,eth2 >> BOOTPROTO=static >> IPADDR=10.0.0.1 >> NETMASK=255.255.255.0 >> >> Obviously, change the necessary parts. >> >> Then run the command: >> >> # /sbin/ifup bond0 >> >> Let us know how this works for you and if you had to take any >> additional steps to get it working. >> >> Greg >> >> >> >> On Wed, Apr 1, 2009 at 12:49 AM, ranjeet.rai at orkash.com >> wrote: >> >>> Hi, >>> >>> I am new to CAOS NSA, looking forward to configure IP bonding. I have >>> configured all the settings and have created the bond0 file with proper >>> entries. This same setting is working fine on CentOS. >>> >>> When I restart the network services get the error message : " I do not >>> understand the TYPE "Ethernet" >>> >>> Please, help me to solve this. >>> >>> Regards, >>> >>> Ranjeet Rai >>> _______________________________________________ >>> cAos mailing list >>> cAos at caoslinux.org >>> http://lists.caosity.org/mailman/listinfo/caos >>> >>> >> >> >> >> > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From stefan at mdy.univie.ac.at Fri Apr 3 00:25:09 2009 From: stefan at mdy.univie.ac.at (Stefan Boresch) Date: Fri, 3 Apr 2009 09:25:09 +0200 Subject: [cAos] mdadm etc. in caos In-Reply-To: <49D514EC.6090306@gmail.com> References: <20090326160716.GL22686@loop.mdy.univie.ac.at> <49D514EC.6090306@gmail.com> Message-ID: <20090403072509.GJ22686@loop.mdy.univie.ac.at> On Thu, Apr 02, 2009 at 12:41:32PM -0700, Greg Kurtzer wrote: > Just wanted to let you know that we didn't miss this email. Thanks for > your patience. Thanks for the info, and thanks for you efforts. I should let you know that I won't be out of town for the next two weeks, so any testing that I could do will have to wait. There is no urgency, since the one machine that is "affected" is up and running by a "hacked-together" rc.local script ... Best regards, Stefan -- Stefan Boresch Institute for Computational Biological Chemistry University of Vienna, Waehringerstr. 17 A-1090 Vienna, Austria Phone: -43-1-427752715 Fax: -43-1-427752790 From glykos at mbg.duth.gr Mon Apr 6 08:27:35 2009 From: glykos at mbg.duth.gr (Nicholas M Glykos) Date: Mon, 6 Apr 2009 18:27:35 +0300 (EEST) Subject: [cAos] telinit 5 ? Message-ID: Dear All, I suspect the answer will be 'google is your friend', but here it goes: On the head node of a CaOS/NSA toy-cluster, I killed X so that I could update the CUDA driver for the GPU. After the driver installation, I tried to restart X by issuing a 'telinit 5' from the console. Unfortunately, this doesn't seem to actually start X. On the other hand, giving a straight 'startx' from the console does indeed start X, resulting to, god forbid, a root's graphics session. What am I missing again ? Thanks, Nicholas -- Dr Nicholas M. Glykos, Department of Molecular Biology and Genetics, Democritus University of Thrace, University Campus, Dragana, 68100 Alexandroupolis, Greece, Tel/Fax (office) +302551030620, Ext.77620, Tel (lab) +302551030615, http://utopia.duth.gr/~glykos/ From gmkurtzer at gmail.com Mon Apr 6 10:33:32 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Mon, 6 Apr 2009 10:33:32 -0700 Subject: [cAos] telinit 5 ? In-Reply-To: References: Message-ID: <571f1a060904061033q5f8ecf96l9c1d8af028fe0f2d@mail.gmail.com> If it is already in run level 5, then you would need to switch out of run level 5 and then back in to activate any changes. but it sounds like you just want to run: $ sudo /etc/init.d/xdm restart Greg On Mon, Apr 6, 2009 at 8:27 AM, Nicholas M Glykos wrote: > > Dear All, > > I suspect the answer will be 'google is your friend', but here it goes: > > On the head node of a CaOS/NSA toy-cluster, I killed X so that I could > update the CUDA driver for the GPU. After the driver installation, I tried > to restart X by issuing a 'telinit 5' from the console. Unfortunately, > this doesn't seem to actually start X. On the other hand, giving a > straight 'startx' from the console does indeed start X, resulting to, god > forbid, a root's graphics session. What am I missing again ? > > > Thanks, > Nicholas > > -- > > > ? ? ? ? ?Dr Nicholas M. Glykos, Department of Molecular Biology > ? ? and Genetics, Democritus University of Thrace, University Campus, > ?Dragana, 68100 Alexandroupolis, Greece, Tel/Fax (office) +302551030620, > ? ?Ext.77620, Tel (lab) +302551030615, http://utopia.duth.gr/~glykos/ > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From glykos at mbg.duth.gr Tue Apr 7 02:24:26 2009 From: glykos at mbg.duth.gr (Nicholas M Glykos) Date: Tue, 7 Apr 2009 12:24:26 +0300 (EEST) Subject: [cAos] telinit 5 ? In-Reply-To: <571f1a060904061033q5f8ecf96l9c1d8af028fe0f2d@mail.gmail.com> References: <571f1a060904061033q5f8ecf96l9c1d8af028fe0f2d@mail.gmail.com> Message-ID: Dear Greg, > $ sudo /etc/init.d/xdm restart Thanks a lot, Nicholas -- Dr Nicholas M. Glykos, Department of Molecular Biology and Genetics, Democritus University of Thrace, University Campus, Dragana, 68100 Alexandroupolis, Greece, Tel/Fax (office) +302551030620, Ext.77620, Tel (lab) +302551030615, http://utopia.duth.gr/~glykos/ From gas5x at yahoo.com Thu Apr 16 07:22:02 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Thu, 16 Apr 2009 07:22:02 -0700 (PDT) Subject: [cAos] manual and automatic disk layouts? Message-ID: <904350.52403.qm@web111306.mail.gq1.yahoo.com> Dear CAOS Users, I'd like to try the Caos-NSA 1 for my small cluster. I've tried to install it -- with defaut layout autoinstall passes OK. But if I want to jave a different layout (for having something like /home on RAID, for exapmle), how do I do that? I've tried layout=manual, but didnt find how to specify where to install Grub, so the system didnt boot after install. So, my question is -- which layouts are available already, and how to install the distribution with manual disk partitioning? Thank you very much! -- With best regards, Grigory Shamov University of Manitoba From gas5x at yahoo.com Thu Apr 16 07:25:33 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Thu, 16 Apr 2009 07:25:33 -0700 (PDT) Subject: [cAos] Installing Torque on Percus Cluster Message-ID: <340087.80461.qm@web111304.mail.gq1.yahoo.com> Dear CAOS Users, Another question -- I think I've seen mentioning that someone installed Torque for Perceus-managed clusters. By default, Sidekick configures SLURM. Is there a way to install Torque, or to access it if it is already in the distro? Thank you! -- With best regards, Grigory Shamov University of Manitoba From astevens at infiscale.com Thu Apr 16 09:09:55 2009 From: astevens at infiscale.com (Arthur Stevens) Date: Thu, 16 Apr 2009 09:09:55 -0700 Subject: [cAos] Installing Torque on Percus Cluster References: <340087.80461.qm@web111304.mail.gq1.yahoo.com> Message-ID: smart install torque torque-server Run this on the server system to install the pbs_server daemon and the command-line utilities for TORQUE. The Caos packaging will automatically update /etc/services with the correct service port entries. It will also pull the system's hostname and populate /var/spool/torque/server_name and /var/spool/torque/server_priv/nodes with it. This sets up a basic default configuration where the server is also its own client. Arthur ----- Original Message ----- From: "Grigory Shamov" To: Sent: Thursday, April 16, 2009 7:25 AM Subject: [cAos] Installing Torque on Percus Cluster > > Dear CAOS Users, > > Another question -- I think I've seen mentioning that someone installed > Torque for Perceus-managed clusters. By default, Sidekick configures > SLURM. Is there a way to install Torque, or to access it if it is already > in the distro? > > Thank you! > > -- > With best regards, > Grigory Shamov > University of Manitoba > > > > > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > From astevens at infiscale.com Thu Apr 16 09:12:13 2009 From: astevens at infiscale.com (Arthur Stevens) Date: Thu, 16 Apr 2009 09:12:13 -0700 Subject: [cAos] manual and automatic disk layouts? References: <904350.52403.qm@web111306.mail.gq1.yahoo.com> Message-ID: <876A97D7F305416B8B0E6043276CCB50@computer> When you do the manual layout, you have to do your own partitioning and mapping. Are you trying to do a software raid or connect to an external raid? Arthur ----- Original Message ----- From: "Grigory Shamov" To: Sent: Thursday, April 16, 2009 7:22 AM Subject: [cAos] manual and automatic disk layouts? > > Dear CAOS Users, > > I'd like to try the Caos-NSA 1 for my small cluster. > I've tried to install it -- with defaut layout autoinstall passes OK. > > But if I want to jave a different layout (for having something like /home > on RAID, for exapmle), how do I do that? I've tried layout=manual, but > didnt find how to specify where to install Grub, so the system didnt boot > after install. > > So, my question is -- which layouts are available already, and how to > install the distribution with manual disk partitioning? > > Thank you very much! > > -- > With best regards, > Grigory Shamov > University of Manitoba > > > > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > From gas5x at yahoo.com Thu Apr 16 11:10:25 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Thu, 16 Apr 2009 11:10:25 -0700 (PDT) Subject: [cAos] manual and automatic disk layouts? Message-ID: <562574.21348.qm@web111311.mail.gq1.yahoo.com> Hi Arthur, Thanks for the reply. Yes, I want to to make a software raid. I've figured out reading this list that I need first partition to be /boot for manual partitioning. If I do it like this, it boots OK. Another question -- I have booted my nodes after the server's installation and upgrade. I use the caos-stateless image, the one that downloads by default. The nodes seem to get added and provisioned by Perceus -- but I cannot log in to them, and they always are in "init", not "ready" state. Are there any hints how to diagnose whats went wrong? -- With best regards, Grigory Shamov University of Manitoba --- On Thu, 4/16/09, Arthur Stevens wrote: > From: Arthur Stevens > Subject: Re: [cAos] manual and automatic disk layouts? > To: "Community-Assembled Operating System" > Date: Thursday, April 16, 2009, 9:12 AM > When you do the manual layout, you > have to do your own partitioning and > mapping. Are you trying to do a software raid or connect to > an external > raid? > > Arthur > > > ----- Original Message ----- > From: "Grigory Shamov" > To: > Sent: Thursday, April 16, 2009 7:22 AM > Subject: [cAos] manual and automatic disk layouts? > > > > > > Dear CAOS Users, > > > > I'd like to try the Caos-NSA 1 for my small cluster. > > I've tried to install it -- with defaut layout > autoinstall passes OK. > > > > But if I want to jave a different layout (for having > something like /home > > on RAID, for exapmle), how do I do that? I've tried > layout=manual, but > > didnt find how to specify where to install Grub, so > the system didnt boot > > after install. > > > > So, my question is -- which layouts are available > already, and how to > > install the distribution with manual disk > partitioning? > > > > Thank you very much! > > > > -- > > With best regards, > > Grigory Shamov > > University of Manitoba > > > > > > > > > > _______________________________________________ > > cAos mailing list > > cAos at caoslinux.org > > http://lists.caosity.org/mailman/listinfo/caos > > > > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > From gmkurtzer at gmail.com Thu Apr 16 12:33:23 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Thu, 16 Apr 2009 12:33:23 -0700 Subject: [cAos] manual and automatic disk layouts? In-Reply-To: <562574.21348.qm@web111311.mail.gq1.yahoo.com> References: <562574.21348.qm@web111311.mail.gq1.yahoo.com> Message-ID: <571f1a060904161233i1b1620b8w15a8734a1a751535@mail.gmail.com> Hello Grigory, Regarding the Perceus node question: Are the nodes network interface getting configured? Can you log into them locally and get a hint as to what is going on? What kind of network interface do they have? Regards, Greg On Thu, Apr 16, 2009 at 11:10 AM, Grigory Shamov wrote: > > Hi Arthur, > > Thanks for the reply. > Yes, I want to to make a software raid. > ?I've figured out reading this list that I need first partition to be /boot for manual partitioning. If I do it like this, it boots OK. > > Another question -- I have booted my nodes after the server's installation and upgrade. I use the caos-stateless image, the one that downloads by default. > > The nodes seem to get added and provisioned by Perceus -- but I cannot log in to them, and they always are in "init", not "ready" state. > > Are there any hints how to diagnose whats went wrong? > > -- > With best regards, > Grigory Shamov > University of Manitoba > > > > > --- On Thu, 4/16/09, Arthur Stevens wrote: > >> From: Arthur Stevens >> Subject: Re: [cAos] manual and automatic disk layouts? >> To: "Community-Assembled Operating System" >> Date: Thursday, April 16, 2009, 9:12 AM >> When you do the manual layout, you >> have to do your own partitioning and >> mapping. Are you trying to do a software raid or connect to >> an external >> raid? >> >> Arthur >> >> >> ----- Original Message ----- >> From: "Grigory Shamov" >> To: >> Sent: Thursday, April 16, 2009 7:22 AM >> Subject: [cAos] manual and automatic disk layouts? >> >> >> > >> > Dear CAOS Users, >> > >> > I'd like to try the Caos-NSA 1 for my small cluster. >> > I've tried to install it -- with defaut layout >> autoinstall passes OK. >> > >> > But if I want to jave a different layout (for having >> something like /home >> > on RAID, for exapmle), how do I do that? I've tried >> layout=manual, but >> > didnt find how to specify where to install Grub, so >> the system didnt boot >> > after install. >> > >> > So, my question is -- which layouts are available >> already, and how to >> > install the distribution with manual disk >> partitioning? >> > >> > Thank you very much! >> > >> > -- >> > With best regards, >> > Grigory Shamov >> > University of Manitoba >> > >> > >> > >> > >> > _______________________________________________ >> > cAos mailing list >> > cAos at caoslinux.org >> > http://lists.caosity.org/mailman/listinfo/caos >> > >> >> >> _______________________________________________ >> cAos mailing list >> cAos at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/caos >> > > > > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From gas5x at yahoo.com Thu Apr 16 13:44:36 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Thu, 16 Apr 2009 13:44:36 -0700 (PDT) Subject: [cAos] manual and automatic disk layouts? Message-ID: <658194.44387.qm@web111309.mail.gq1.yahoo.com> Hi Greg, I cannot log in locally. The nodes seem to get OS from the server, then freese for a long time at the stage of mounting NFS. After that, they show login prompt but root cannot login (always wrong password). I cannot ping or ssh to them from the server either. I have Sun x2200 servers, with two eth0, eth1 cabled for the nodes (the latter I've planned to use as "fast" network for MPI traffic). For the server, eth1 serves the outer network connection. Unplugging eth1's from the switch during installation didnt help somehow (i have deleted nodes from perceus, then left only eth0 plugged and rediscovered them again). --- On Thu, 4/16/09, Greg Kurtzer wrote: > From: Greg Kurtzer > Subject: Re: [cAos] manual and automatic disk layouts? > To: "Community-Assembled Operating System" > Date: Thursday, April 16, 2009, 12:33 PM > Hello Grigory, > > Regarding the Perceus node question: > > Are the nodes network interface getting configured? > > Can you log into them locally and get a hint as to what is > going on? > > What kind of network interface do they have? > > > Regards, > Greg > > > On Thu, Apr 16, 2009 at 11:10 AM, Grigory Shamov > wrote: > > > > Hi Arthur, > > > > Thanks for the reply. > > Yes, I want to to make a software raid. > > ?I've figured out reading this list that I need first > partition to be /boot for manual partitioning. If I do it > like this, it boots OK. > > > > Another question -- I have booted my nodes after the > server's installation and upgrade. I use the caos-stateless > image, the one that downloads by default. > > > > The nodes seem to get added and provisioned by Perceus > -- but I cannot log in to them, and they always are in > "init", not "ready" state. > > > > Are there any hints how to diagnose whats went wrong? > > > > -- > > With best regards, > > Grigory Shamov > > University of Manitoba > > > > From astevens at Infiscale.com Thu Apr 16 14:42:09 2009 From: astevens at Infiscale.com (Arthur A. Stevens) Date: Thu, 16 Apr 2009 14:42:09 -0700 Subject: [cAos] manual and automatic disk layouts? In-Reply-To: <562574.21348.qm@web111311.mail.gq1.yahoo.com> References: <562574.21348.qm@web111311.mail.gq1.yahoo.com> Message-ID: <49E7A631.3090603@Infiscale.com> If your systems have the nvidia forcedeath issue, then a smart update && smart upgrade and reboot should fix that up if using an older ISO than -current. Let me know if that helps, Arthur Grigory Shamov wrote: > Hi Arthur, > > Thanks for the reply. > Yes, I want to to make a software raid. > I've figured out reading this list that I need first partition to be /boot for manual partitioning. If I do it like this, it boots OK. > > Another question -- I have booted my nodes after the server's installation and upgrade. I use the caos-stateless image, the one that downloads by default. > > The nodes seem to get added and provisioned by Perceus -- but I cannot log in to them, and they always are in "init", not "ready" state. > > Are there any hints how to diagnose whats went wrong? > > -- > With best regards, > Grigory Shamov > University of Manitoba > > > > > --- On Thu, 4/16/09, Arthur Stevens wrote: > > >> From: Arthur Stevens >> Subject: Re: [cAos] manual and automatic disk layouts? >> To: "Community-Assembled Operating System" >> Date: Thursday, April 16, 2009, 9:12 AM >> When you do the manual layout, you >> have to do your own partitioning and >> mapping. Are you trying to do a software raid or connect to >> an external >> raid? >> >> Arthur >> >> >> ----- Original Message ----- >> From: "Grigory Shamov" >> To: >> Sent: Thursday, April 16, 2009 7:22 AM >> Subject: [cAos] manual and automatic disk layouts? >> >> >> >>> Dear CAOS Users, >>> >>> I'd like to try the Caos-NSA 1 for my small cluster. >>> I've tried to install it -- with defaut layout >>> >> autoinstall passes OK. >> >>> But if I want to jave a different layout (for having >>> >> something like /home >> >>> on RAID, for exapmle), how do I do that? I've tried >>> >> layout=manual, but >> >>> didnt find how to specify where to install Grub, so >>> >> the system didnt boot >> >>> after install. >>> >>> So, my question is -- which layouts are available >>> >> already, and how to >> >>> install the distribution with manual disk >>> >> partitioning? >> >>> Thank you very much! >>> >>> -- >>> With best regards, >>> Grigory Shamov >>> University of Manitoba >>> >>> >>> >>> >>> _______________________________________________ >>> cAos mailing list >>> cAos at caoslinux.org >>> http://lists.caosity.org/mailman/listinfo/caos >>> >>> >> _______________________________________________ >> cAos mailing list >> cAos at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/caos >> >> > > > > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > From gas5x at yahoo.com Thu Apr 16 15:27:35 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Thu, 16 Apr 2009 15:27:35 -0700 (PDT) Subject: [cAos] manual and automatic disk layouts? Message-ID: <891595.89088.qm@web111313.mail.gq1.yahoo.com> Hi Arthur, I used the current (1.0.8) Caos NSA iso and did update it after install. For my systems, the NICs I use are broadcom tg3, i.e. first two. There are two nvidia forcedeth -- supposed to be disabled; I didnt plug the cables in them. Despite that somehow on the nodes in the output it says at some point that it is going to provision from eth2. But it proceeds so I think it works via eth0 somehow, at least before it tries to mount NFS volumes? -- With best regards, Grigory Shamov --- On Thu, 4/16/09, Arthur A. Stevens wrote: > From: Arthur A. Stevens > Subject: Re: [cAos] manual and automatic disk layouts? > To: "Community-Assembled Operating System" > Date: Thursday, April 16, 2009, 2:42 PM > If your systems have the nvidia > forcedeath issue, then a smart update && > smart upgrade and reboot should fix that up if using an > older ISO than > -current. Let me know if that helps, > > Arthur > > Grigory Shamov wrote: > > Hi Arthur, > > > > Thanks for the reply. > > Yes, I want to to make a software raid. > >? I've figured out reading this list that I need > first partition to be /boot for manual partitioning. If I do > it like this, it boots OK. > > > > Another question -- I have booted my nodes after the > server's installation and upgrade. I use the caos-stateless > image, the one that downloads by default. > > > > The nodes seem to get added and provisioned by Perceus > -- but I cannot log in to them, and they always are in > "init", not "ready" state.. > > > > Are there any hints how to diagnose whats went wrong? > > > > -- > > With best regards, > > Grigory Shamov > > University of Manitoba > > > > > > > > > > --- On Thu, 4/16/09, Arthur Stevens > wrote: > > > >??? > >> From: Arthur Stevens > >> Subject: Re: [cAos] manual and automatic disk > layouts? > >> To: "Community-Assembled Operating System" > >> Date: Thursday, April 16, 2009, 9:12 AM > >> When you do the manual layout, you > >> have to do your own partitioning and > >> mapping. Are you trying to do a software raid or > connect to > >> an external > >> raid? > >> > >> Arthur > >> > >> > >> ----- Original Message ----- > >> From: "Grigory Shamov" > >> To: > >> Sent: Thursday, April 16, 2009 7:22 AM > >> Subject: [cAos] manual and automatic disk > layouts? > >> > >> > >>? ??? > >>> Dear CAOS Users, > >>> > >>> I'd like to try the Caos-NSA 1 for my small > cluster. > >>> I've tried to install it -- with defaut > layout > >>>? ? ??? > >> autoinstall passes OK. > >>? ??? > >>> But if I want to jave a different layout (for > having > >>>? ? ??? > >> something like /home > >>? ??? > >>> on RAID, for exapmle), how do I do that? I've > tried > >>>? ? ??? > >> layout=manual, but > >>? ??? > >>> didnt find how to specify where to install > Grub, so > >>>? ? ??? > >> the system didnt boot > >>? ??? > >>> after install. > >>> > >>> So, my question is -- which layouts are > available > >>>? ? ??? > >> already, and how to > >>? ??? > >>> install the distribution with manual disk > >>>? ? ??? > >> partitioning? > >>? ??? > >>> Thank you very much! > >>> > >>> -- > >>> With best regards, > >>> Grigory Shamov > >>> University of Manitoba > >>> > >>> > >>> > >>> > >>> > _______________________________________________ > >>> cAos mailing list > >>> cAos at caoslinux.org > >>> http://lists.caosity.org/mailman/listinfo/caos > >>> > >>>? ? ??? > >> _______________________________________________ > >> cAos mailing list > >> cAos at caoslinux..org > >> http://lists.caosity.org/mailman/listinfo/caos > >> > >>? ??? > > > > > >? ? ??? > > > > _______________________________________________ > > cAos mailing list > > cAos at caoslinux..org > > http://lists.caosity.org/mailman/listinfo/caos > >??? > > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > From gmkurtzer at gmail.com Thu Apr 16 16:54:07 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Thu, 16 Apr 2009 16:54:07 -0700 Subject: [cAos] manual and automatic disk layouts? In-Reply-To: <891595.89088.qm@web111313.mail.gq1.yahoo.com> References: <891595.89088.qm@web111313.mail.gq1.yahoo.com> Message-ID: <571f1a060904161654g46603ab0ue0976df25ace126d@mail.gmail.com> Hello Grigory, You can always add eth2 to the /etc/perceus/modules/ipaddr config file (it currently sets a eth0 and eth1). To debug this I would recommend to log in locally at the console of the node and see what is getting configured, and try and isolate what is going on. Good luck, Greg On Thu, Apr 16, 2009 at 3:27 PM, Grigory Shamov wrote: > > Hi Arthur, > > I used the current (1.0.8) Caos NSA iso and did update it after install. > > For my systems, the NICs I use are broadcom tg3, i.e. first two. There are two nvidia forcedeth -- supposed to be disabled; I didnt plug the cables in them. > > Despite that somehow on the nodes in the output it says at some point that it is going to provision from eth2. But it ?proceeds so I think it works via eth0 somehow, at least before it tries to mount NFS volumes? > > -- > With best regards, > Grigory Shamov > > --- On Thu, 4/16/09, Arthur A. Stevens wrote: > >> From: Arthur A. Stevens >> Subject: Re: [cAos] manual and automatic disk layouts? >> To: "Community-Assembled Operating System" >> Date: Thursday, April 16, 2009, 2:42 PM >> If your systems have the nvidia >> forcedeath issue, then a smart update && >> smart upgrade and reboot should fix that up if using an >> older ISO than >> -current. Let me know if that helps, >> >> Arthur >> >> Grigory Shamov wrote: >> > Hi Arthur, >> > >> > Thanks for the reply. >> > Yes, I want to to make a software raid. >> >? I've figured out reading this list that I need >> first partition to be /boot for manual partitioning. If I do >> it like this, it boots OK. >> > >> > Another question -- I have booted my nodes after the >> server's installation and upgrade. I use the caos-stateless >> image, the one that downloads by default. >> > >> > The nodes seem to get added and provisioned by Perceus >> -- but I cannot log in to them, and they always are in >> "init", not "ready" state.. >> > >> > Are there any hints how to diagnose whats went wrong? >> > >> > -- >> > With best regards, >> > Grigory Shamov >> > University of Manitoba >> > >> > >> > >> > >> > --- On Thu, 4/16/09, Arthur Stevens >> wrote: >> > >> > >> >> From: Arthur Stevens >> >> Subject: Re: [cAos] manual and automatic disk >> layouts? >> >> To: "Community-Assembled Operating System" >> >> Date: Thursday, April 16, 2009, 9:12 AM >> >> When you do the manual layout, you >> >> have to do your own partitioning and >> >> mapping. Are you trying to do a software raid or >> connect to >> >> an external >> >> raid? >> >> >> >> Arthur >> >> >> >> >> >> ----- Original Message ----- >> >> From: "Grigory Shamov" >> >> To: >> >> Sent: Thursday, April 16, 2009 7:22 AM >> >> Subject: [cAos] manual and automatic disk >> layouts? >> >> >> >> >> >> >> >>> Dear CAOS Users, >> >>> >> >>> I'd like to try the Caos-NSA 1 for my small >> cluster. >> >>> I've tried to install it -- with defaut >> layout >> >>> >> >> autoinstall passes OK. >> >> >> >>> But if I want to jave a different layout (for >> having >> >>> >> >> something like /home >> >> >> >>> on RAID, for exapmle), how do I do that? I've >> tried >> >>> >> >> layout=manual, but >> >> >> >>> didnt find how to specify where to install >> Grub, so >> >>> >> >> the system didnt boot >> >> >> >>> after install. >> >>> >> >>> So, my question is -- which layouts are >> available >> >>> >> >> already, and how to >> >> >> >>> install the distribution with manual disk >> >>> >> >> partitioning? >> >> >> >>> Thank you very much! >> >>> >> >>> -- >> >>> With best regards, >> >>> Grigory Shamov >> >>> University of Manitoba >> >>> >> >>> >> >>> >> >>> >> >>> >> _______________________________________________ >> >>> cAos mailing list >> >>> cAos at caoslinux.org >> >>> http://lists.caosity.org/mailman/listinfo/caos >> >>> >> >>> >> >> _______________________________________________ >> >> cAos mailing list >> >> cAos at caoslinux..org >> >> http://lists.caosity.org/mailman/listinfo/caos >> >> >> >> >> > >> > >> > >> > >> > _______________________________________________ >> > cAos mailing list >> > cAos at caoslinux..org >> > http://lists.caosity.org/mailman/listinfo/caos >> > >> >> >> _______________________________________________ >> cAos mailing list >> cAos at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/caos >> > > > > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From gas5x at yahoo.com Fri Apr 17 13:03:14 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Fri, 17 Apr 2009 13:03:14 -0700 (PDT) Subject: [cAos] hosts on nodes and DHCP on server Message-ID: <198135.32281.qm@web111306.mail.gq1.yahoo.com> Hi Greg, Thank you! Adding debug helped. I've found that during the node provisioning, tg3 and forcedeth devices of my Sun x2200's change places, and due to that NFS mount stalls. After I switched the nvidia NICs off in BIOS, all my nodes booted to the "ready" state succesfully. I can log in to them from the server. Now, the booted nodes do not see each other (there is nothing in hosts but localhost). Could you please tell, how do I enable proper host files or DNS? Another question: there is a Management Console on x2200's eth1. It can take IP's from Perceus's DHCP server, which is quite handy, but could be also confusing with new nodes. Is there a way to specify MC's by their MACs as MCs, that is, have their IPs given but not processed as nodes? Thank you very much! -- With best ragards, Grigory Shamov Chemistry University of Manitoba --- On Thu, 4/16/09, Greg Kurtzer wrote: > From: Greg Kurtzer > Subject: Re: [cAos] manual and automatic disk layouts? > To: "Community-Assembled Operating System" > Date: Thursday, April 16, 2009, 4:54 PM > Hello Grigory, > > You can always add eth2 to the /etc/perceus/modules/ipaddr > config file > (it currently sets a eth0 and eth1). > > To debug this I would recommend to log in locally at the > console of > the node and see what is getting configured, and try and > isolate what > is going on. > > Good luck, > Greg From gmkurtzer at gmail.com Fri Apr 17 13:40:32 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Fri, 17 Apr 2009 13:40:32 -0700 Subject: [cAos] hosts on nodes and DHCP on server In-Reply-To: <198135.32281.qm@web111306.mail.gq1.yahoo.com> References: <198135.32281.qm@web111306.mail.gq1.yahoo.com> Message-ID: <571f1a060904171340k478de3f0kac347b6e24e9b94c@mail.gmail.com> Try to set the node's /etc/resolv.conf to point at the master node's IP address. Edit it directly on a node to test, and then you can make the change persistent by: # perceus vnfs mount [vnfs name] # vi /mnt/[vnfs name]/etc/resolv.conf # perceus vnfs umount [vnfs name] You can use the Perceus DHCP server with no issue about Perceus confusing the IP's with nodes. If you want Perceus to assign static IP addresses in its DHCP configuration, then edit /etc/perceus/dnsmasq.conf and add something like: dhcp-host=11:22:33:44:55:66,10.0.0.100 Hope that helps! On Fri, Apr 17, 2009 at 1:03 PM, Grigory Shamov wrote: > > Hi Greg, > > Thank you! Adding debug helped. > > I've found that during the node provisioning, tg3 and forcedeth devices of my Sun x2200's change places, and due to that NFS mount stalls. After I switched the nvidia NICs off in BIOS, all my nodes booted to the "ready" state succesfully. I can log in to them from the server. > > Now, the booted nodes do not see each other (there is nothing in hosts but localhost). Could you please tell, how do I enable proper host files or DNS? > > Another question: there is a Management Console on x2200's eth1. It can take IP's from Perceus's DHCP server, which is quite handy, but could be also confusing with new nodes. Is there a way to specify MC's by their MACs as MCs, that is, have their IPs given but not processed as nodes? > > Thank you very much! > > -- > With best ragards, > Grigory Shamov > Chemistry > University of Manitoba > > > > > --- On Thu, 4/16/09, Greg Kurtzer wrote: > >> From: Greg Kurtzer >> Subject: Re: [cAos] manual and automatic disk layouts? >> To: "Community-Assembled Operating System" >> Date: Thursday, April 16, 2009, 4:54 PM >> Hello Grigory, >> >> You can always add eth2 to the /etc/perceus/modules/ipaddr >> config file >> (it currently sets a eth0 and eth1). >> >> To debug this I would recommend to log in locally at the >> console of >> the node and see what is getting configured, and try and >> isolate what >> is going on. >> >> Good luck, >> Greg > > > > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From gas5x at yahoo.com Sun Apr 19 12:24:26 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Sun, 19 Apr 2009 12:24:26 -0700 (PDT) Subject: [cAos] raids on local disks - hybridization? Message-ID: <987932.89099.qm@web111315.mail.gq1.yahoo.com> Dear All, Some more questions (sorry if there are too many -- my previous experience was with statful clusters like OSCAR mostly). 1) So, I have my nodes booting my Caos vnfs capsule etc, and can do some alterations in the capsules, like, adding packages there using Smart. I can mount partitions on local HDDs, which I need as a fast scratch space for user's applications. Since I have two HDDs per node, I want to make RAID0 array of them, for speed. I can add mdadm package to my capsule, log in to node, create the RAID0 and it works OK. But how do make a capsule to access it? I mean, when there is no /dev/mdX device and no mdadm configs in the capsule, and it doesnt get stored after reboot? So, could you please provide your suggestions and/or share experience with RAID access from the stateless nodes? 2) I need my nodes have more software (like, Python with Numeric, Perl, etc.), but not bloat its RAM usage. I've read that there is a mechanizm called Hybridization to provide parts of system from the server. But the documentation is rather cryptic -- could you provide an example -- say, if I want my /usr "hybridized", what is the sequence of steps? Thank you! -- With best regards, Grigory Shamov Chemistry, University of Manitoba From ranjeet.rai at orkash.com Mon Apr 20 04:04:29 2009 From: ranjeet.rai at orkash.com (ranjeet.rai at orkash.com) Date: Mon, 20 Apr 2009 16:34:29 +0530 Subject: [cAos] configuring diskless in CAOS Message-ID: <49EC56BD.2090105@orkash.com> Hi, Can anybody guide me on configuring diskless environment in CAOS NSA? Best regards, Ranjeet Rai From darr at ocean.washington.edu Mon Apr 20 09:56:29 2009 From: darr at ocean.washington.edu (david darr) Date: Mon, 20 Apr 2009 09:56:29 -0700 Subject: [cAos] raids on local disks - hybridization? In-Reply-To: <987932.89099.qm@web111315.mail.gq1.yahoo.com> References: <987932.89099.qm@web111315.mail.gq1.yahoo.com> Message-ID: <6a8af7e80904200956m51e33544k2672a97e635f24af@mail.gmail.com> > 2) I need my nodes have more software (like, Python with Numeric, Perl, > etc.), but not bloat its RAM usage. I've read that there is a mechanizm > called Hybridization to provide parts of system from the server. But the > documentation is rather cryptic -- could you provide an example -- say, if I > want my /usr "hybridized", what is the sequence of steps? Thank you! Grigory, The way I do this is to just NFS mount from the master node. In particular, I mount /usr/local from the master node to /usr/local in the vnfs capsule and put all of the additional software there. Works fine. There may be other ways to do it but this works well for my needs. In particular, here is the entry from /etc/fstab on one compute node. 10.1.1.1:/usr/local /usr/local nfs defaults 0 0 David -- -------------------------------------------------------------------------- David Darr - ph: (206) 616-4953 - email: darr at ocean.washington.edu School of Oceanography - Univ. of Washington - Seattle, WA. 98195 -------------------------------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.infiscale.org/pipermail/caos/attachments/20090420/a8752e6c/attachment.html From gas5x at yahoo.com Mon Apr 20 12:11:32 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Mon, 20 Apr 2009 12:11:32 -0700 (PDT) Subject: [cAos] raids on local disks - hybridization? Message-ID: <422681.49309.qm@web111313.mail.gq1.yahoo.com> Hi David, How do you deal with /var? It should be rw, and it can be useful to see logs on nodes after reboot. But, obviously one cannot just export one /var for all the nodes for rw? -- With best regards, Grigory Shamov Chemistry, University of Manitoba --- On Mon, 4/20/09, david darr wrote: > From: david darr > Subject: Re: [cAos] raids on local disks - hybridization? > To: "Community-Assembled Operating System" > Date: Monday, April 20, 2009, 9:56 AM > > > 2) I need my nodes have more software (like, Python with > Numeric, Perl, etc.), but not bloat its RAM usage. > ?I've read that there is a mechanizm called > Hybridization to provide parts of system from the server. > But the documentation is rather cryptic -- could you provide > an example -- say, if I want my /usr "hybridized", > what is the sequence of steps? Thank you! > > Grigory, > > The way I do this is to just NFS mount from the master > node. In particular, I mount /usr/local from the master node > to /usr/local in the vnfs capsule and put all of the > additional software there. Works fine. There may be other > ways to do it but this works well for my needs. > > > In particular, here is the entry from /etc/fstab on one > compute node. > > 10.1.1.1:/usr/local??????? > /usr/local?????????? nfs defaults 0 0 > > David > > -- > > -------------------------------------------------------------------------- > > ?David Darr - ph: (206) 616-4953 - email: darr at ocean.washington.edu > School of Oceanography - Univ. of Washington - Seattle, WA. > 98195 > -------------------------------------------------------------------------- > > > > > > -----Inline Attachment Follows----- > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > From gas5x at yahoo.com Mon Apr 20 12:26:45 2009 From: gas5x at yahoo.com (gas5x at yahoo.com) Date: Mon, 20 Apr 2009 12:26:45 -0700 (PDT) Subject: [cAos] configuring diskless in CAOS Message-ID: <275771.48502.qm@web111304.mail.gq1.yahoo.com> Hi Ranjeet, To get started, there is a walkthrough, on the Caos NSA website. http://wiki.caoslinux.org/Caos_NSA_1_Installation_Walkthrough . I've followed it: installed the Caos NSA1 on server, configured Perceus for the Computational Caos-NSA capsule, then booted my nodes with PXE boot enabled. One can monitor the progress by "perceus node status" command -- watch your nodes appear in the table. If you get this working, then you start customizing your capsules to your needs -- mounting them by perceus vnfs mount, altering, then umounting and rebooting your nodes. -- With best regards, Grigory Shamov Chemistry, University of Manitoba --- On Mon, 4/20/09, ranjeet.rai at orkash.com wrote: > From: ranjeet.rai at orkash.com > Subject: [cAos] configuring diskless in CAOS > To: caos at caoslinux.org > Date: Monday, April 20, 2009, 4:04 AM > Hi, > > Can anybody guide me on configuring diskless environment in > CAOS NSA? > > Best regards, > > Ranjeet Rai > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > From darr at ocean.washington.edu Mon Apr 20 12:29:26 2009 From: darr at ocean.washington.edu (david darr) Date: Mon, 20 Apr 2009 12:29:26 -0700 Subject: [cAos] raids on local disks - hybridization? In-Reply-To: <422681.49309.qm@web111313.mail.gq1.yahoo.com> References: <422681.49309.qm@web111313.mail.gq1.yahoo.com> Message-ID: <6a8af7e80904201229x1e471542l66e394a9c4a4c636@mail.gmail.com> > How do you deal with /var? > It should be rw, and it can be useful to see logs on nodes after reboot. > But, obviously one cannot just export one /var for all the nodes for rw? That's a good question and I don't have a good answer for it. I've wondered about this myself. The Perceus default is apparently to write back to the logs in /var on the master node but I have not verified how (or even if) this works. Hopefully someone else has the answer..... -- -------------------------------------------------------------------------- David Darr - ph: (206) 616-4953 - email: darr at ocean.washington.edu School of Oceanography - Univ. of Washington - Seattle, WA. 98195 -------------------------------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.infiscale.org/pipermail/caos/attachments/20090420/a5086c36/attachment.html From glykos at mbg.duth.gr Mon Apr 20 13:12:27 2009 From: glykos at mbg.duth.gr (Nicholas M Glykos) Date: Mon, 20 Apr 2009 23:12:27 +0300 (EEST) Subject: [cAos] raids on local disks - hybridization? In-Reply-To: <422681.49309.qm@web111313.mail.gq1.yahoo.com> References: <422681.49309.qm@web111313.mail.gq1.yahoo.com> Message-ID: Hi Grigory, > How do you deal with /var? On one of our (stateless) nodes as installed using the defaults: # # cat /etc/fstab # Don't touch the following lines unless you really know what you're doing! none / tmpfs defaults 0 0 none /dev/pts devpts gid=5,mode=620 0 0 none /dev/shm tmpfs defaults 0 0 none /proc proc defaults 0 0 none /tmp tmpfs size=256M 0 0 # To use a local disk on the nodes (make sure the mountpoints exist in # the VNFS!!!) #/dev/hda2 /scratch ext3 defaults 0 0 #/dev/hda1 none swap defaults 0 0 10.0.0.1:/var/lib/perceus/ /var/lib/perceus/ nfs ro,soft,bg 0 0 10.0.0.1:/home/ /home nfs defaults 0 0 10.0.0.1:/opt /opt nfs defaults 0 0 10.0.0.1:/usr/local /usr/local nfs defaults 0 0 10.0.0.1:/usr/cports /usr/cports nfs defaults 0 0 10.0.0.1:/srv /srv nfs defaults 0 0 10.0.0.1:/private /private nfs defaults 0 0 > It should be rw, and it can be useful to see logs on nodes after reboot. There are practically no logs to see. On the same node: # # cd /var/log # ls -lF total 36 -rw-r--r-- 1 root root 21525 Apr 14 12:01 dmesg -rw-r--r-- 1 root root 3601 Apr 20 19:34 ntp.log -rw-rw-r-- 1 root utmp 7680 Apr 20 23:03 wtmp # To my mind this is just as it should: the node is just a pluggable box containing 'CPU+memory+network'. Having to do log management on a per node basis would have defeated the purpose. My twopence, Nicholas -- Dr Nicholas M. Glykos, Department of Molecular Biology and Genetics, Democritus University of Thrace, University Campus, 68100 Alexandroupolis, Greece, Fax +302551030613 Tel ++302551030620 (77620), http://www.mbg.duth.gr/~glykos/ From gas5x at yahoo.com Mon Apr 20 15:48:48 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Mon, 20 Apr 2009 15:48:48 -0700 (PDT) Subject: [cAos] compiling Maui on Caos NSA Linux: mktime problem? Message-ID: <100997.88415.qm@web111304.mail.gq1.yahoo.com> Dear All, I've tried to build Maui from sources, on my newly installed Caos NSA v.1.0.8 Linux. It has a new version of GCC (here is the output of gcc -v) ==== Using built-in specs. Target: x86_64-caos-linux-gnu Configured with: ../configure --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --libdir=/usr/lib64 --libexecdir=/usr/lib64 --with-slib=/lib64 --infodir=/usr/share/info --mandir=/usr/share/man --enable-shared --enable-threads=posix --with-gxx-include-dir=/usr/include/c++/4.3.3 --enable-languages=c,c++,fortran --disable-checking --enable-nls --enable-c-mbchar --enable-long-long --enable-__cxa_atexit --host=x86_64-caos-linux-gnu --build=x86_64-caos-linux-gnu --target=x86_64-caos-linux-gnu Thread model: posix gcc version 4.3.3 (GCC) ==== When I do the Maui's configure script, it freezed at the following point: ==== .... checking for GNU libc compatible malloc... yes checking for working memcmp... yes checking for sys/time.h... (cached) yes checking for unistd.h... (cached) yes checking for alarm... no checking for working mktime... ^C ==== Could you please tell how to make it work? Do I need any additional packages to have this mktime thing "working"? Thank you! -- With best regards, Grigory Shamov, Chemistry University of Manitoba From mej at caoslinux.org Mon Apr 20 20:41:51 2009 From: mej at caoslinux.org (Michael Jennings) Date: Mon, 20 Apr 2009 20:41:51 -0700 Subject: [cAos] compiling Maui on Caos NSA Linux: mktime problem? In-Reply-To: <100997.88415.qm@web111304.mail.gq1.yahoo.com> References: <100997.88415.qm@web111304.mail.gq1.yahoo.com> Message-ID: <20090421034151.GA4844@kainx.org> On Monday, 20 April 2009, at 15:48:48 (-0700), Grigory Shamov wrote: > When I do the Maui's configure script, it freezed at the following point: > > ==== > .... > checking for GNU libc compatible malloc... yes > checking for working memcmp... yes > checking for sys/time.h... (cached) yes > checking for unistd.h... (cached) yes > checking for alarm... no > checking for working mktime... ^C > ==== > > Could you please tell how to make it work? Do I need any additional packages to have this mktime thing "working"? mktime() is not the problem. The problem is that conftest, a program generated and executed by the configure script, is hung. Kill it, and configure will continue to completion. Michael -- Michael Jennings (a.k.a. KainX) http://www.kainx.org/ Linux Server/Cluster Admin, LBL.gov Author, Eterm (www.eterm.org) ----------------------------------------------------------------------- "Well done is better than well said." -- Benjamin Franklin From gas5x at yahoo.com Tue Apr 21 10:57:20 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Tue, 21 Apr 2009 10:57:20 -0700 (PDT) Subject: [cAos] raid again - passing options to kernel? Message-ID: <691583.20554.qm@web111306.mail.gq1.yahoo.com> Dear Caos Users, It seems that for my local scratch RAID to get autodetected at boot time, kernel options need to be passed. Could you please tell, how to manage kernels for the perceus nodes -- i.e., passing them arguments, picking different kernels and so on? Thank you! -- With best regards, Grigory Shamov Chemistry, University of Manitoba From gas5x at yahoo.com Tue Apr 21 13:37:36 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Tue, 21 Apr 2009 13:37:36 -0700 (PDT) Subject: [cAos] user management for Caos perceus nodes? Message-ID: <324591.91384.qm@web111316.mail.gq1.yahoo.com> Dear Caos-Users, How do you add users to be visible on the stateless nodes? I understand that a module called "passwdfile" should take care of providing passwd/shadow entries to the nodes, and keys for each new user generate at his first login. I've added a new user by plain useradd, and ssh keys for him were generated. But somehow the user's entry is not in the nodes' passwd (and not in /etc/perceus/modules/passwdfile/all file as well), so he cant login. Do I have to reboot nodes after each change in my users? Or I have to edit the module's file manually to enable them? -- With best regards, Grigory Shamov Chemistry, University of Manitoba From darr at ocean.washington.edu Tue Apr 21 14:00:41 2009 From: darr at ocean.washington.edu (david darr) Date: Tue, 21 Apr 2009 14:00:41 -0700 Subject: [cAos] user management for Caos perceus nodes? In-Reply-To: <324591.91384.qm@web111316.mail.gq1.yahoo.com> References: <324591.91384.qm@web111316.mail.gq1.yahoo.com> Message-ID: <6a8af7e80904211400i6466a6b2i87b34a1d039f8d3@mail.gmail.com> > How do you add users to be visible on the stateless nodes? > > I understand that a module called "passwdfile" should take care of providing passwd/shadow entries to the nodes, and keys for each new user generate at his first login. > > I've added a new user by plain useradd, and ssh keys for him were generated. But somehow the user's entry is not in the nodes' passwd (and not in /etc/perceus/modules/passwdfile/all file as well), so he cant login. > > Do I have to reboot nodes after each change in my users? Or I have to edit the module's file manually to enable them? Hello Grigory, Here is what worked for me. This was a bit trial and error for me as the documentation was a bit sketchy. Make sure that you have the module passswd enabled. Do "perceus module summary" -- it should give you: passwdfile: ready/all Then, in /etc/perceus/modules/passwdfile, you should have several directories and the file "all". "all" is the passwd file that you want communicated to your nodes (I just cp /etc/passwd from the master node to this every time I update and add/delete a user) and no, you don't have to reboot the nodes. The nodes automatically check in and update their /etc/passwd to match this file. This last bit caught me by surprise in fact and caused quite a bit of head-scratching at first before I realized what was happening. The same holds true for the groupfile module as well. -David -- -------------------------------------------------------------------------- David Darr - ph: (206) 616-4953 - email: darr at ocean.washington.edu School of Oceanography - Univ. of Washington - Seattle, WA. 98195 -------------------------------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.infiscale.org/pipermail/caos/attachments/20090421/9acb944d/attachment.html From gmkurtzer at gmail.com Tue Apr 21 14:33:40 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Tue, 21 Apr 2009 14:33:40 -0700 Subject: [cAos] user management for Caos perceus nodes? In-Reply-To: <6a8af7e80904211400i6466a6b2i87b34a1d039f8d3@mail.gmail.com> References: <324591.91384.qm@web111316.mail.gq1.yahoo.com> <6a8af7e80904211400i6466a6b2i87b34a1d039f8d3@mail.gmail.com> Message-ID: <571f1a060904211433q4d84263bw50ed3f7bb922834a@mail.gmail.com> Another module to look at is the "masterauth" perceus module. It syncronizes the user list from the master node and grants all users access to the nodes (assuming correct ssh keys exist). Caos NSA already makes these for the user when they first log in, so if your using NSA, just activate the module, and within several minutes all nodes will have gotten updated and ready for logins. Gerg On Tue, Apr 21, 2009 at 2:00 PM, david darr wrote: >> How do you add users to be visible on the stateless nodes? >> >> I understand that a module called "passwdfile" should take care of >> providing passwd/shadow entries to the nodes, and keys for each new user >> ?generate at his first login. >> >> I've added a new user by plain useradd, and ssh keys for him were >> generated. But somehow the user's entry is not in the nodes' passwd (and not >> in ?/etc/perceus/modules/passwdfile/all file as well), so he cant login. >> >> Do I have to reboot nodes after each change in my users? Or I have to edit >> the module's file manually to enable them? > > Hello Grigory, > > Here is what worked for me. This was a bit trial and error for me as the > documentation was a bit sketchy. > > > Make sure that you have the module passswd enabled. Do "perceus module > summary"? -- it should give you: > > passwdfile: > ??????? ready/all > > Then, in /etc/perceus/modules/passwdfile, you should have several > directories and the file "all". "all" is the passwd file that you want > communicated to your nodes (I just cp /etc/passwd from the master node to > this every time I update and add/delete a user) and no, you don't have to > reboot the nodes. The nodes automatically check in and update their > /etc/passwd to match this file. This last bit caught me by surprise in fact > and caused quite a bit of head-scratching at first before I realized what > was happening. > > The same holds true for the groupfile module as well. > > -David > > > -- > > -------------------------------------------------------------------------- > ?David Darr - ph: (206) 616-4953 - email: darr at ocean.washington.edu > School of Oceanography - Univ. of Washington - Seattle, WA. 98195 > -------------------------------------------------------------------------- > > > > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From gmkurtzer at gmail.com Tue Apr 21 14:35:28 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Tue, 21 Apr 2009 14:35:28 -0700 Subject: [cAos] raid again - passing options to kernel? In-Reply-To: <691583.20554.qm@web111306.mail.gq1.yahoo.com> References: <691583.20554.qm@web111306.mail.gq1.yahoo.com> Message-ID: <571f1a060904211435o3f71107dq16fd8589bca84468@mail.gmail.com> Check out /etc/perceus/vnfs/[vnfs name]/config. :) Greg On Tue, Apr 21, 2009 at 10:57 AM, Grigory Shamov wrote: > > Dear Caos Users, > > It seems that for my local scratch RAID to get autodetected at boot time, kernel options need to be passed. > > Could you please tell, how to manage kernels for the perceus nodes -- i.e., passing them arguments, picking different kernels and so on? > Thank you! > > -- > With best regards, > Grigory Shamov > Chemistry, > University of Manitoba > > > > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From jscoggins at lbl.gov Tue Apr 21 14:36:20 2009 From: jscoggins at lbl.gov (Jackie Scoggins) Date: Tue, 21 Apr 2009 14:36:20 -0700 Subject: [cAos] raid again - passing options to kernel? In-Reply-To: <571f1a060904211435o3f71107dq16fd8589bca84468@mail.gmail.com> References: <691583.20554.qm@web111306.mail.gq1.yahoo.com> <571f1a060904211435o3f71107dq16fd8589bca84468@mail.gmail.com> Message-ID: <441D1518-45D5-49F9-A6CE-52A593AC331E@lbl.gov> R u n your office On Apr 21, 2009, at 2:35 PM, Greg Kurtzer wrote: > Check out /etc/perceus/vnfs/[vnfs name]/config. :) > > Greg > > > > On Tue, Apr 21, 2009 at 10:57 AM, Grigory Shamov > wrote: >> >> Dear Caos Users, >> >> It seems that for my local scratch RAID to get autodetected at boot >> time, kernel options need to be passed. >> >> Could you please tell, how to manage kernels for the perceus nodes >> -- i.e., passing them arguments, picking different kernels and so on? >> Thank you! >> >> -- >> With best regards, >> Grigory Shamov >> Chemistry, >> University of Manitoba >> >> >> >> >> _______________________________________________ >> cAos mailing list >> cAos at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/caos >> > > > > -- > Greg Kurtzer > http://www.infiscale.com/ > http://www.perceus.org/ > http://www.caoslinux.org/ > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos From gas5x at yahoo.com Tue Apr 21 14:50:26 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Tue, 21 Apr 2009 14:50:26 -0700 (PDT) Subject: [cAos] raid again - passing options to kernel? Message-ID: <666388.91384.qm@web111303.mail.gq1.yahoo.com> Thank you! Found it! :-) --- On Tue, 4/21/09, Greg Kurtzer wrote: > From: Greg Kurtzer > Subject: Re: [cAos] raid again - passing options to kernel? > To: "Community-Assembled Operating System" > Date: Tuesday, April 21, 2009, 2:35 PM > Check out /etc/perceus/vnfs/[vnfs > name]/config. :) > > Greg > > > > On Tue, Apr 21, 2009 at 10:57 AM, Grigory Shamov > wrote: > > > > Dear Caos Users, > > > > It seems that for my local scratch RAID to get > autodetected at boot time, kernel options need to be > passed. > > > > Could you please tell, how to manage kernels for the > perceus nodes -- i.e., passing them arguments, picking > different kernels and so on? > > Thank you! > > > > -- > > With best regards, > > Grigory Shamov > > Chemistry, > > University of Manitoba > > > > > > > > > > _______________________________________________ > > cAos mailing list > > cAos at caoslinux.org > > http://lists.caosity.org/mailman/listinfo/caos > > > > > > -- > Greg Kurtzer > http://www.infiscale.com/ > http://www.perceus.org/ > http://www.caoslinux.org/ > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > From gas5x at yahoo.com Tue Apr 21 14:50:49 2009 From: gas5x at yahoo.com (Grigory Shamov) Date: Tue, 21 Apr 2009 14:50:49 -0700 (PDT) Subject: [cAos] user management for Caos perceus nodes? Message-ID: <450485.13649.qm@web111304.mail.gq1.yahoo.com> Thank you very much! Now it works! --- On Tue, 4/21/09, david darr wrote: > From: david darr > Subject: Re: [cAos] user management for Caos perceus nodes? > To: "Community-Assembled Operating System" > Date: Tuesday, April 21, 2009, 2:00 PM > > How do you add users to be > visible on the stateless nodes? > > > > I understand that a module called > "passwdfile" should take care of providing > passwd/shadow entries to the nodes, and keys for each new > user ?generate at his first login. > > > > > I've added a new user by plain useradd, and ssh > keys for him were generated. But somehow the user's > entry is not in the nodes' passwd (and not in > ?/etc/perceus/modules/passwdfile/all file as well), so he > cant login. > > > > > Do I have to reboot nodes after each change in my > users? Or I have to edit the module's file manually to > enable them? > > Hello Grigory, > > Here is what worked for me. This was a bit trial and error > for me as the documentation was a bit sketchy. > > > > Make sure that you have the module passswd enabled. Do > "perceus module summary"? -- it should give you: > > passwdfile: > ??????? ready/all > > Then, in /etc/perceus/modules/passwdfile, you should have > several directories and the file "all". > "all" is the passwd file that you want > communicated to your nodes (I just cp /etc/passwd from the > master node to this every time I update and add/delete a > user) and no, you don't have to reboot the nodes. The > nodes automatically check in and update their /etc/passwd to > match this file. This last bit caught me by surprise in fact > and caused quite a bit of head-scratching at first before I > realized what was happening. > > > The same holds true for the groupfile module as well. > > -David > > > -- > > -------------------------------------------------------------------------- > ?David Darr - ph: (206) 616-4953 - email: darr at ocean.washington.edu > > School of Oceanography - Univ. of Washington - Seattle, WA. > 98195 > -------------------------------------------------------------------------- > > > > > > -----Inline Attachment Follows----- > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > From kirkby at etsu.edu Fri Apr 24 07:30:46 2009 From: kirkby at etsu.edu (Scott Kirkby) Date: Fri, 24 Apr 2009 10:30:46 -0400 Subject: [cAos] Installing caos-nsa on HP DL385 with Smart Array Controller 6i Message-ID: <49F1CD16.7020400@etsu.edu> Hi, Has anyone installed caos-nsa 1.08 on an HP DL385 with a HP Smart Array controller 6i? Cinch finds the controller and loads the cciss module but then is unable to find the logical volume to complete the install. I have tried disk=cciss/c0d0 but this does not work and there is no cciss/c0d0 entry in /dev running in test mode. Thanks, Scott From gas5x at yahoo.com Tue Apr 28 07:44:43 2009 From: gas5x at yahoo.com (gas5x at yahoo.com) Date: Tue, 28 Apr 2009 07:44:43 -0700 (PDT) Subject: [cAos] enabling root ssh login from node to node Message-ID: <207664.54384.qm@web111303.mail.gq1.yahoo.com> Dear Caos-Users, I've found that my root account can only log in from the server to each diskless node, but not from one node to another. For regular users ssh does work (since their keys are sitting on the shared /home/$USER/.ssh). Could anyone suggest how to enable root's node to node ssh login? Thank you! -- With best regards, Grigory Shamov Chemistry University of Manitoba From gmkurtzer at gmail.com Tue Apr 28 14:56:02 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Tue, 28 Apr 2009 14:56:02 -0700 Subject: [cAos] enabling root ssh login from node to node In-Reply-To: <207664.54384.qm@web111303.mail.gq1.yahoo.com> References: <207664.54384.qm@web111303.mail.gq1.yahoo.com> Message-ID: <571f1a060904281456w677f31eey652c094e47461b15@mail.gmail.com> Hello Grigory, You would need to incorporate root's public and private ssh keys on the node VNFS image then. I would very strongly recommend NOT to do this for security reasons. Thanks, Greg On Tue, Apr 28, 2009 at 7:44 AM, wrote: > > Dear Caos-Users, > > I've found that my root account can only log in from the server to each diskless node, but not from one node to another. > > For regular users ssh does work (since their keys are sitting on the shared /home/$USER/.ssh). > > Could anyone suggest how to enable root's node to node ssh login? Thank you! > > -- > With best regards, > Grigory Shamov > Chemistry > University of Manitoba > > > > > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From gmkurtzer at gmail.com Tue Apr 28 15:00:49 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Tue, 28 Apr 2009 15:00:49 -0700 Subject: [cAos] Installing caos-nsa on HP DL385 with Smart Array Controller 6i In-Reply-To: <49F1CD16.7020400@etsu.edu> References: <49F1CD16.7020400@etsu.edu> Message-ID: <571f1a060904281500p3be9d9c9ubbde056ee5defb59@mail.gmail.com> Hello Scott, Recently we have had some people working on this in current cinch, as well as a new installer revision in development that would make this easier to support. Our developers don't have a system with this hardware, so fixing up cinch to support it is a bit tricky. Are you available to test a new version when we get it released or maybe help us add support for the cciss? Thanks, Greg On Fri, Apr 24, 2009 at 7:30 AM, Scott Kirkby wrote: > Hi, > > Has anyone installed caos-nsa 1.08 on an HP DL385 with a HP Smart Array > controller 6i? > > Cinch finds the controller and loads the cciss module but then is unable > to find the logical volume to complete the install. I have tried > disk=cciss/c0d0 but this does not work and there is no cciss/c0d0 entry > in /dev running in test mode. > > Thanks, > > Scott > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From kirkby at etsu.edu Tue Apr 28 15:43:48 2009 From: kirkby at etsu.edu (Scott Kirkby) Date: Tue, 28 Apr 2009 18:43:48 -0400 Subject: [cAos] Installing caos-nsa on HP DL385 with Smart Array Controller 6i In-Reply-To: <571f1a060904281500p3be9d9c9ubbde056ee5defb59@mail.gmail.com> References: <49F1CD16.7020400@etsu.edu> <571f1a060904281500p3be9d9c9ubbde056ee5defb59@mail.gmail.com> Message-ID: <49F786A4.8010608@etsu.edu> Greg, I would be willing to test it and do what I can. I do not have any device driver experience. I am just a chemist that uses Linux to do my quantum mechanics calculations. An additional complication. I can get caos-nsa to install on a different SCSI disk controller on the DL385. It will then recognize the cciss disk array. I can mount the logical volume manually but the entry in /etc/fstab is ignored and I can not find any error messages in the logs. Thanks, Scott Greg Kurtzer wrote: > Hello Scott, > > Recently we have had some people working on this in current cinch, as > well as a new installer revision in development that would make this > easier to support. > > Our developers don't have a system with this hardware, so fixing up > cinch to support it is a bit tricky. > > Are you available to test a new version when we get it released or > maybe help us add support for the cciss? > > Thanks, > Greg > > > > On Fri, Apr 24, 2009 at 7:30 AM, Scott Kirkby wrote: > >> Hi, >> >> Has anyone installed caos-nsa 1.08 on an HP DL385 with a HP Smart Array >> controller 6i? >> >> Cinch finds the controller and loads the cciss module but then is unable >> to find the logical volume to complete the install. I have tried >> disk=cciss/c0d0 but this does not work and there is no cciss/c0d0 entry >> in /dev running in test mode. >> >> Thanks, >> >> Scott >> >> _______________________________________________ >> cAos mailing list >> cAos at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/caos >> >> > > > > From gmkurtzer at gmail.com Tue Apr 28 17:04:12 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Tue, 28 Apr 2009 17:04:12 -0700 Subject: [cAos] Installing caos-nsa on HP DL385 with Smart Array Controller 6i In-Reply-To: <49F786A4.8010608@etsu.edu> References: <49F1CD16.7020400@etsu.edu> <571f1a060904281500p3be9d9c9ubbde056ee5defb59@mail.gmail.com> <49F786A4.8010608@etsu.edu> Message-ID: <571f1a060904281704l707da558x513b663b5e5a2526@mail.gmail.com> Scott, Caos itself will/should recognize the cciss disk array just fine, but the installer doesn't have the logic to deal with it yet. Hopefully that will change soon. ;) What is the entry in /etc/fstab for the cciss device? Does the device exist at boot (/dev/cciss/*)? What if you type "mount -a" after it boots? Regards, Greg On Tue, Apr 28, 2009 at 3:43 PM, Scott Kirkby wrote: > Greg, > > I would be willing to test it and do what I can. I do not have any > device driver experience. I am just a chemist that uses Linux to do my > quantum mechanics calculations. > > An additional complication. I can get caos-nsa to install on a different > SCSI disk controller on the DL385. It will then recognize the cciss disk > array. I can mount the logical volume manually but the entry in > /etc/fstab is ignored and I can not find any error messages in the logs. > > Thanks, > > Scott > > Greg Kurtzer wrote: >> Hello Scott, >> >> Recently we have had some people working on this in current cinch, as >> well as a new installer revision in development that would make this >> easier to support. >> >> Our developers don't have a system with this hardware, so fixing up >> cinch to support it is a bit tricky. >> >> Are you available to test a new version when we get it released or >> maybe help us add support for the cciss? >> >> Thanks, >> Greg >> >> >> >> On Fri, Apr 24, 2009 at 7:30 AM, Scott Kirkby wrote: >> >>> Hi, >>> >>> Has anyone installed caos-nsa 1.08 on an HP DL385 with a HP Smart Array >>> controller 6i? >>> >>> Cinch finds the controller and loads the cciss module but then is unable >>> to find the logical volume to complete the install. I have tried >>> disk=cciss/c0d0 but this does not work and there is no cciss/c0d0 entry >>> in /dev running in test mode. >>> >>> Thanks, >>> >>> Scott >>> >>> _______________________________________________ >>> cAos mailing list >>> cAos at caoslinux.org >>> http://lists.caosity.org/mailman/listinfo/caos >>> >>> >> >> >> >> > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From caos at commobil.de Wed Apr 29 02:36:00 2009 From: caos at commobil.de (Mike Bobkiewicz) Date: Wed, 29 Apr 2009 11:36:00 +0200 Subject: [cAos] Problem with process viewing Message-ID: <8C78CCA6-F6B9-482C-8EAB-C04EF4CF8707@commobil.de> Hi, while I was installing the Monitoring software Zabbix I stumbled into a problem with Caos Linux: a user only can see his own processes. While Zabbix refuses to start with any other user than "zabbix" it can only see the processes he has started. And while one of the main goals of Zabbix is to inform you about running processes the whole system becomes pretty obsolete. Where can I change this behaviour so any user on the system can see all running processes on the system? Greetings, Mike From gmkurtzer at gmail.com Wed Apr 29 06:51:42 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Wed, 29 Apr 2009 06:51:42 -0700 Subject: [cAos] Problem with process viewing In-Reply-To: <8C78CCA6-F6B9-482C-8EAB-C04EF4CF8707@commobil.de> References: <8C78CCA6-F6B9-482C-8EAB-C04EF4CF8707@commobil.de> Message-ID: <571f1a060904290651h214a2669o4893410923af7c72@mail.gmail.com> Hello Mike, Please see the following page describing and disabling the proc security for any user(s) in Caos Linux: http://wiki.caoslinux.org/Secadm If you have any questions please let us know! Greg On Wed, Apr 29, 2009 at 2:36 AM, Mike Bobkiewicz wrote: > Hi, > while I was installing the Monitoring software Zabbix I stumbled into > a problem with Caos Linux: a user only can see his own processes. > While Zabbix refuses to start with any other user than "zabbix" it can > only see the processes he has started. And while one of the main goals > of Zabbix is to inform you about running processes the whole system > becomes pretty obsolete. Where can I change this behaviour so any user > on the system can see all running processes on the system? > > Greetings, > > ? Mike > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From kirkby at etsu.edu Wed Apr 29 13:38:05 2009 From: kirkby at etsu.edu (Scott Kirkby) Date: Wed, 29 Apr 2009 16:38:05 -0400 Subject: [cAos] Installing caos-nsa on HP DL385 with Smart Array Controller 6i In-Reply-To: <571f1a060904281704l707da558x513b663b5e5a2526@mail.gmail.com> References: <49F1CD16.7020400@etsu.edu> <571f1a060904281500p3be9d9c9ubbde056ee5defb59@mail.gmail.com> <49F786A4.8010608@etsu.edu> <571f1a060904281704l707da558x513b663b5e5a2526@mail.gmail.com> Message-ID: <49F8BAAD.6070805@etsu.edu> Greg, Fstab is: # FSTAB created by Cinch (the NSA installer) /dev/sda3 / ext4 defaults 1 1 /dev/sda1 /boot ext3 defaults 1 2 /dev/sda2 none swap defaults 0 0 /dev/cciss/c0d0p1 /array ext4 defaults 1 3 none /tmp tmpfs size=512m 0 0 rpc_pipefs /var/lib/nfs/rpc_pipefs rpc_pipefs defaults 0 0 and the output from mount -a is mount: rpc_pipefs already mounted or /var/lib/nfs/rpc_pipefs busy mount: according to mtab, sunrpc is already mounted on /var/lib/nfs/rpc_pipefs Thanks, Scott Greg Kurtzer wrote: > Scott, > > Caos itself will/should recognize the cciss disk array just fine, but > the installer doesn't have the logic to deal with it yet. Hopefully > that will change soon. ;) > > What is the entry in /etc/fstab for the cciss device? Does the device > exist at boot (/dev/cciss/*)? What if you type "mount -a" after it > boots? > > Regards, > Greg > > On Tue, Apr 28, 2009 at 3:43 PM, Scott Kirkby wrote: > >> Greg, >> >> I would be willing to test it and do what I can. I do not have any >> device driver experience. I am just a chemist that uses Linux to do my >> quantum mechanics calculations. >> >> An additional complication. I can get caos-nsa to install on a different >> SCSI disk controller on the DL385. It will then recognize the cciss disk >> array. I can mount the logical volume manually but the entry in >> /etc/fstab is ignored and I can not find any error messages in the logs. >> >> Thanks, >> >> Scott >> >> Greg Kurtzer wrote: >> >>> Hello Scott, >>> >>> Recently we have had some people working on this in current cinch, as >>> well as a new installer revision in development that would make this >>> easier to support. >>> >>> Our developers don't have a system with this hardware, so fixing up >>> cinch to support it is a bit tricky. >>> >>> Are you available to test a new version when we get it released or >>> maybe help us add support for the cciss? >>> >>> Thanks, >>> Greg >>> >>> >>> >>> On Fri, Apr 24, 2009 at 7:30 AM, Scott Kirkby wrote: >>> >>> >>>> Hi, >>>> >>>> Has anyone installed caos-nsa 1.08 on an HP DL385 with a HP Smart Array >>>> controller 6i? >>>> >>>> Cinch finds the controller and loads the cciss module but then is unable >>>> to find the logical volume to complete the install. I have tried >>>> disk=cciss/c0d0 but this does not work and there is no cciss/c0d0 entry >>>> in /dev running in test mode. >>>> >>>> Thanks, >>>> >>>> Scott >>>> >>>> _______________________________________________ >>>> cAos mailing list >>>> cAos at caoslinux.org >>>> http://lists.caosity.org/mailman/listinfo/caos >>>> >>>> >>>> >>> >>> >>> >> _______________________________________________ >> cAos mailing list >> cAos at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/caos >> >> > > > > From gmkurtzer at gmail.com Wed Apr 29 14:10:29 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Wed, 29 Apr 2009 14:10:29 -0700 Subject: [cAos] Installing caos-nsa on HP DL385 with Smart Array Controller 6i In-Reply-To: <49F8BAAD.6070805@etsu.edu> References: <49F1CD16.7020400@etsu.edu> <571f1a060904281500p3be9d9c9ubbde056ee5defb59@mail.gmail.com> <49F786A4.8010608@etsu.edu> <571f1a060904281704l707da558x513b663b5e5a2526@mail.gmail.com> <49F8BAAD.6070805@etsu.edu> Message-ID: <571f1a060904291410r41759089i419cca94b7a5681d@mail.gmail.com> Hello Scott, I think we know what is causing this. Please update to sysinit-1.2.9-1.caos which is available in the nsa-testing repository, and then reboot. For information on how to configure the nsa-testing repository, please see: http://wiki.caoslinux.org/Activating_nsa-testing_repo_with_Sidekick Let us know how this works please. Greg On Wed, Apr 29, 2009 at 1:38 PM, Scott Kirkby wrote: > Greg, > > Fstab is: > > # FSTAB created by Cinch (the NSA installer) > > /dev/sda3 ? ? ? / ? ? ? ext4 ? ?defaults ? ? ? ?1 1 > /dev/sda1 ? ? ? /boot ? ext3 ? ?defaults ? ? ? ?1 2 > /dev/sda2 ? ? ? none ? ?swap ? ?defaults ? ? ? ?0 0 > /dev/cciss/c0d0p1 ? ? ? /array ?ext4 ? ?defaults ? ? ? ?1 3 > > none ? ? ? /tmp ? ?tmpfs ? size=512m ? ? ? 0 0 > rpc_pipefs /var/lib/nfs/rpc_pipefs rpc_pipefs defaults 0 0 > > > and the output from mount -a is > > mount: rpc_pipefs already mounted or /var/lib/nfs/rpc_pipefs busy > mount: according to mtab, sunrpc is already mounted on > /var/lib/nfs/rpc_pipefs > > Thanks, > > Scott > > > Greg Kurtzer wrote: >> Scott, >> >> Caos itself will/should recognize the cciss disk array just fine, but >> the installer doesn't have the logic to deal with it yet. Hopefully >> that will change soon. ;) >> >> What is the entry in /etc/fstab for the cciss device? Does the device >> exist at boot (/dev/cciss/*)? What if you type "mount -a" after it >> boots? >> >> Regards, >> Greg >> >> On Tue, Apr 28, 2009 at 3:43 PM, Scott Kirkby wrote: >> >>> Greg, >>> >>> I would be willing to test it and do what I can. I do not have any >>> device driver experience. I am just a chemist that uses Linux to do my >>> quantum mechanics calculations. >>> >>> An additional complication. I can get caos-nsa to install on a different >>> SCSI disk controller on the DL385. It will then recognize the cciss disk >>> array. I can mount the logical volume manually but the entry in >>> /etc/fstab is ignored and I can not find any error messages in the logs. >>> >>> Thanks, >>> >>> Scott >>> >>> Greg Kurtzer wrote: >>> >>>> Hello Scott, >>>> >>>> Recently we have had some people working on this in current cinch, as >>>> well as a new installer revision in development that would make this >>>> easier to support. >>>> >>>> Our developers don't have a system with this hardware, so fixing up >>>> cinch to support it is a bit tricky. >>>> >>>> Are you available to test a new version when we get it released or >>>> maybe help us add support for the cciss? >>>> >>>> Thanks, >>>> Greg >>>> >>>> >>>> >>>> On Fri, Apr 24, 2009 at 7:30 AM, Scott Kirkby wrote: >>>> >>>> >>>>> Hi, >>>>> >>>>> Has anyone installed caos-nsa 1.08 on an HP DL385 with a HP Smart Array >>>>> controller 6i? >>>>> >>>>> Cinch finds the controller and loads the cciss module but then is unable >>>>> to find the logical volume to complete the install. I have tried >>>>> disk=cciss/c0d0 but this does not work and there is no cciss/c0d0 entry >>>>> in /dev running in test mode. >>>>> >>>>> Thanks, >>>>> >>>>> Scott >>>>> >>>>> _______________________________________________ >>>>> cAos mailing list >>>>> cAos at caoslinux.org >>>>> http://lists.caosity.org/mailman/listinfo/caos >>>>> >>>>> >>>>> >>>> >>>> >>>> >>> _______________________________________________ >>> cAos mailing list >>> cAos at caoslinux.org >>> http://lists.caosity.org/mailman/listinfo/caos >>> >>> >> >> >> >> > > _______________________________________________ > cAos mailing list > cAos at caoslinux.org > http://lists.caosity.org/mailman/listinfo/caos > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From kirkby at etsu.edu Thu Apr 30 15:52:58 2009 From: kirkby at etsu.edu (Scott Kirkby) Date: Thu, 30 Apr 2009 18:52:58 -0400 Subject: [cAos] Installing caos-nsa on HP DL385 with Smart Array Controller 6i In-Reply-To: <571f1a060904291410r41759089i419cca94b7a5681d@mail.gmail.com> References: <49F1CD16.7020400@etsu.edu> <571f1a060904281500p3be9d9c9ubbde056ee5defb59@mail.gmail.com> <49F786A4.8010608@etsu.edu> <571f1a060904281704l707da558x513b663b5e5a2526@mail.gmail.com> <49F8BAAD.6070805@etsu.edu> <571f1a060904291410r41759089i419cca94b7a5681d@mail.gmail.com> Message-ID: <49FA2BCA.9020000@etsu.edu> Hi Greg, After installing sysinit-1.2.9-1.caos the system mounts the drive on booting without any problems. Thanks, Scott Greg Kurtzer wrote: > Hello Scott, > > I think we know what is causing this. Please update to > sysinit-1.2.9-1.caos which is available in the nsa-testing repository, > and then reboot. For information on how to configure the nsa-testing > repository, please see: > > http://wiki.caoslinux.org/Activating_nsa-testing_repo_with_Sidekick > > Let us know how this works please. > > Greg > > > On Wed, Apr 29, 2009 at 1:38 PM, Scott Kirkby wrote: > >> Greg, >> >> Fstab is: >> >> # FSTAB created by Cinch (the NSA installer) >> >> /dev/sda3 / ext4 defaults 1 1 >> /dev/sda1 /boot ext3 defaults 1 2 >> /dev/sda2 none swap defaults 0 0 >> /dev/cciss/c0d0p1 /array ext4 defaults 1 3 >> >> none /tmp tmpfs size=512m 0 0 >> rpc_pipefs /var/lib/nfs/rpc_pipefs rpc_pipefs defaults 0 0 >> >> >> and the output from mount -a is >> >> mount: rpc_pipefs already mounted or /var/lib/nfs/rpc_pipefs busy >> mount: according to mtab, sunrpc is already mounted on >> /var/lib/nfs/rpc_pipefs >> >> Thanks, >> >> Scott >> >> >> Greg Kurtzer wrote: >> >>> Scott, >>> >>> Caos itself will/should recognize the cciss disk array just fine, but >>> the installer doesn't have the logic to deal with it yet. Hopefully >>> that will change soon. ;) >>> >>> What is the entry in /etc/fstab for the cciss device? Does the device >>> exist at boot (/dev/cciss/*)? What if you type "mount -a" after it >>> boots? >>> >>> Regards, >>> Greg >>> >>> On Tue, Apr 28, 2009 at 3:43 PM, Scott Kirkby wrote: >>> >>> >>>> Greg, >>>> >>>> I would be willing to test it and do what I can. I do not have any >>>> device driver experience. I am just a chemist that uses Linux to do my >>>> quantum mechanics calculations. >>>> >>>> An additional complication. I can get caos-nsa to install on a different >>>> SCSI disk controller on the DL385. It will then recognize the cciss disk >>>> array. I can mount the logical volume manually but the entry in >>>> /etc/fstab is ignored and I can not find any error messages in the logs. >>>> >>>> Thanks, >>>> >>>> Scott >>>> >>>> Greg Kurtzer wrote: >>>> >>>> >>>>> Hello Scott, >>>>> >>>>> Recently we have had some people working on this in current cinch, as >>>>> well as a new installer revision in development that would make this >>>>> easier to support. >>>>> >>>>> Our developers don't have a system with this hardware, so fixing up >>>>> cinch to support it is a bit tricky. >>>>> >>>>> Are you available to test a new version when we get it released or >>>>> maybe help us add support for the cciss? >>>>> >>>>> Thanks, >>>>> Greg >>>>> >>>>> >>>>> >>>>> On Fri, Apr 24, 2009 at 7:30 AM, Scott Kirkby wrote: >>>>> >>>>> >>>>> >>>>>> Hi, >>>>>> >>>>>> Has anyone installed caos-nsa 1.08 on an HP DL385 with a HP Smart Array >>>>>> controller 6i? >>>>>> >>>>>> Cinch finds the controller and loads the cciss module but then is unable >>>>>> to find the logical volume to complete the install. I have tried >>>>>> disk=cciss/c0d0 but this does not work and there is no cciss/c0d0 entry >>>>>> in /dev running in test mode. >>>>>> >>>>>> Thanks, >>>>>> >>>>>> Scott >>>>>> >>>>>> _______________________________________________ >>>>>> cAos mailing list >>>>>> cAos at caoslinux.org >>>>>> http://lists.caosity.org/mailman/listinfo/caos >>>>>> >>>>>> >>>>>> >>>>>> >>>>> >>>>> >>>> _______________________________________________ >>>> cAos mailing list >>>> cAos at caoslinux.org >>>> http://lists.caosity.org/mailman/listinfo/caos >>>> >>>> >>>> >>> >>> >>> >> _______________________________________________ >> cAos mailing list >> cAos at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/caos >> >> > > > >