Menu
▾
▴
xcat-user
|
From: bfulda at 3drc.c. (B. Fulda) - 2005-06-10 09:50:24
|
Hey all, I have now a functioning xcat server. I still have a few bugs to work out, like wcon doesn't work and so on, but one thing I need to fix is the SW that is installed. When a node is installed I only get cli. I need everything. All three disks of RH9. Where do I make this change? I see in the /install directory there is a file on down that sets up things like type of install, how to set up the disks, etc. But what is the nomenclature for an "ALL" SW load. Tia Brad Fulda 256-651-5949 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.xcat.org/pipermail/xcat-user/attachments/20050610/dd4e92bc/attachment.htm |
|
From: bfulda at 3drc.c. (B. Fulda) - 2005-06-10 10:36:26
|
You might be right. After I inserted "@Everything" into /install/rh9/x86/compute file. I get the blue screen and "waiting on telnet connection" and it stays there waiting. RH9 has the ability to load "Everything" I read. SUSE doesn't. Maybe I'll try deleting the other @ stuff and just leave the @Everything. At least before I was getting a load, now I'm hanging. Tia for any help Brad ________________________________ From: Scott Denham [mailto:sd...@us...] Sent: Friday, June 10, 2005 11:21 AM To: xCAT Users Mailing list Cc: Brad Fulda Subject: Re: [xcat-user] Software loads Brad - This file is documented in the Red Hat System Administration Guide. I know the installer has a checkbox for "Everything", but I don't find a comparable item in the comps.xml file. Maybe you just need to grep all the groups out of there and insert them. In doing so, I'll bet you find some stuff you don't *really* want on your cluster nodes. Yes, disk is cheap, but the more stuff you install, the longer it takes :) Here's a helpful extract regarding package selection: Packages can be specified by group or by individual package name. The installation program defines several groups that contain related packages. Refer to the RedHat/base/comps.xml file on the first Red Hat Enterprise Linux CD-ROM for a list of groups. Each group has an id, user visibility value, name, description, and package list. In the package list, the packages marked as mandatory are always installed if the group is selected, the packages marked default are selected by default if the group is selected, and the packages marked optional must be specifically selected even if the group is selected to be installed. ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- || Scott S. Denham || HPC Technical Architect || Petroleum Lead - Deep Computing Technical Team || IBM Systems & Technology Group || Two Riverway Email: sd...@us... || Houston, TX 77056 Voice/Fax: 713.940.1178 || (CST / GMT - 6) ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- xca...@li... wrote on 06/10/2005 10:50:24 AM: > Hey all, > > I have now a functioning xcat server. I still have a few bugs to > work out, like wcon doesn't work and so on, but one thing I need to > fix is the SW that is installed. When a node is installed I only > get cli. I need everything. All three disks of RH9. Where do I > make this change? I see in the /install directory there is a file on > down that sets up things like type of install, how to set up the > disks, etc. But what is the nomenclature for an "ALL" SW load. Tia > Brad Fulda > 256-651-5949_______________________________________________ > xcat-user mailing list > xca...@li... > http://www.xcat.org/mailman/listinfo/xcat-user -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.xcat.org/pipermail/xcat-user/attachments/20050610/1eb81dce/attachment-0001.htm |
|
From: egan at sense.n. (E. Ford) - 2005-06-12 22:58:37
|
Did you use: nodename:|conserver.telnet nodename::&: in $XCATROOT/etc/conserver.cf? That is the only reason I can think of that would have RH9 install prompting for telnet. This is used for remote console access to machines that do not have terminal servers or SOL, but that you want to still monitor the install. -----Original Message----- From: xca...@li... [mailto:xca...@li...] On Behalf Of Brad Fulda Sent: Friday, June 10, 2005 10:36 AM To: Scott Denham; xCAT Users Mailing list Subject: RE: [xcat-user] Software loads Importance: High You might be right. After I inserted "@Everything" into /install/rh9/x86/compute file. I get the blue screen and "waiting on telnet connection" and it stays there waiting. RH9 has the ability to load "Everything" I read. SUSE doesn't. Maybe I'll try deleting the other @ stuff and just leave the @Everything. At least before I was getting a load, now I'm hanging. Tia for any help Brad _____ From: Scott Denham [mailto:sd...@us...] Sent: Friday, June 10, 2005 11:21 AM To: xCAT Users Mailing list Cc: Brad Fulda Subject: Re: [xcat-user] Software loads Brad - This file is documented in the Red Hat System Administration Guide. I know the installer has a checkbox for "Everything", but I don't find a comparable item in the comps.xml file. Maybe you just need to grep all the groups out of there and insert them. In doing so, I'll bet you find some stuff you don't *really* want on your cluster nodes. Yes, disk is cheap, but the more stuff you install, the longer it takes :) Here's a helpful extract regarding package selection: Packages can be specified by group or by individual package name. The installation program defines several groups that contain related packages. Refer to the RedHat/base/comps.xml file on the first Red Hat Enterprise Linux CD-ROM for a list of groups. Each group has an id, user visibility value, name, description, and package list. In the package list, the packages marked as mandatory are always installed if the group is selected, the packages marked default are selected by default if the group is selected, and the packages marked optional must be specifically selected even if the group is selected to be installed. ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- || Scott S. Denham || HPC Technical Architect || Petroleum Lead - Deep Computing Technical Team || IBM Systems & Technology Group || Two Riverway Email: sd...@us... || Houston, TX 77056 Voice/Fax: 713.940.1178 || (CST / GMT - 6) ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- xca...@li... wrote on 06/10/2005 10:50:24 AM: > Hey all, > > I have now a functioning xcat server. I still have a few bugs to > work out, like wcon doesn't work and so on, but one thing I need to > fix is the SW that is installed. When a node is installed I only > get cli. I need everything. All three disks of RH9. Where do I > make this change? I see in the /install directory there is a file on > down that sets up things like type of install, how to set up the > disks, etc. But what is the nomenclature for an "ALL" SW load. Tia > Brad Fulda > 256-651-5949_______________________________________________ > xcat-user mailing list > xca...@li... > http://www.xcat.org/mailman/listinfo/xcat-user -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.xcat.org/pipermail/xcat-user/attachments/20050612/3b026bee/attachment.htm |
|
From: mabugaev at unity.ncsu.e. (M. Bugaev) - 2005-06-13 14:42:06
|
Hi all, Have any of you seen the following behavior:? BladeCenter (BC) chassis has Ethernet Switch Module (ESM) in bay1 and Copper Pass-thru Module (CPM) in bay2. Blade 8843-E2U in chassis would like to be loaded with RHEL3 via CPM (eth1). All necessary settings are there, I believe, like "eth1" in noderes.tab. However after getting address from DHCPD and loading install kernel, it fails to get address for eth1 second time during kickstart procedure and goes to interactive install. And BTW, if I specify IP address it supposed to get from DHCP by hand, I'm able to mount install directory from head-node and continue interactive install. If I put this blade in the chassis with ESM in both bays, everything goes as expected. What am I missing here? Thanks in advance for any suggestions, Regards, Michael |
|
From: sdenham at us.ibm.c. (S. Denham) - 2005-06-13 16:50:00
|
Skipped content of type multipart/alternative-------------- next part -------------- A non-text attachment was scrubbed... Name: graycol.gif Type: image/gif Size: 105 bytes Desc: not available Url : http://www.xcat.org/pipermail/xcat-user/attachments/20050613/a192262d/graycol.gif -------------- next part -------------- A non-text attachment was scrubbed... Name: pic13213.gif Type: image/gif Size: 1255 bytes Desc: not available Url : http://www.xcat.org/pipermail/xcat-user/attachments/20050613/a192262d/pic13213.gif -------------- next part -------------- A non-text attachment was scrubbed... Name: ecblank.gif Type: image/gif Size: 45 bytes Desc: not available Url : http://www.xcat.org/pipermail/xcat-user/attachments/20050613/a192262d/ecblank.gif |
|
From: mabugaev at unity.ncsu.e. (M. A Bugaev) - 2005-06-14 05:20:25
|
Hi Scott, Both chassis are connected to Cisco 6509. I'm not sure about spanning tree on it, it's not under my control, need to ask network engineer. But why it does not happen with second chassis, even thought there is extra switch between blade and head-node comparing to chassis with CPM? Doesn't it make that one more likely to suffer long delays? In any case, thanks for an idea. Regards, Michael On Mon, 13 Jun 2005, Scott Denham wrote: > > > > > > What's the copper pass thru plugged into? Does it have spanning-tree > protocol disabled? The long delays incurred by spanning tree analyzing the > network for loops is known to cause this timeout in the RH installer.... > > > ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- > =- > || Scott S. Denham > || HPC Technical Architect > || Petroleum Lead - Deep Computing Technical Team > || IBM Systems & Technology Group > || Two Riverway Email: sd...@us... > || Houston, TX 77056 Voice/Fax: 713.940.1178 > || (CST / > GMT - 6) > ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- > =- > > > > > Michael Bugaev > <mabugaev@unity.n > csu.edu> To > Sent by: xCAT Users Mailing list > xcat-user-bounces <xca...@li...> > @lists.xcat.org cc > > Subject > 06/13/2005 03:42 [xcat-user] BladeCenter with Copper > PM Pass-thru module > > > Please respond to > xCAT Users > Mailing list > > > > > > > Hi all, > > Have any of you seen the following behavior:? > > BladeCenter (BC) chassis has Ethernet Switch Module (ESM) in bay1 and > Copper Pass-thru Module (CPM) in bay2. Blade 8843-E2U in chassis would > like to be loaded with RHEL3 via CPM (eth1). All necessary settings are > there, I believe, like "eth1" in noderes.tab. However after getting > address from DHCPD and loading install kernel, it fails to get address > for eth1 second time during kickstart procedure and goes to interactive > install. And BTW, if I specify IP address it supposed to get from DHCP > by hand, I'm able to mount install directory from head-node and continue > interactive install. > If I put this blade in the chassis with ESM in both bays, everything > goes as expected. What am I missing here? > > Thanks in advance for any suggestions, > Regards, > Michael > _______________________________________________ > xcat-user mailing list > xca...@li... > http://www.xcat.org/mailman/listinfo/xcat-user > |
|
From: egan at sense.n. (E. Ford) - 2005-06-14 08:25:20
|
The links between the two switches are already established when blades are in an ESM connected to a Cisco. With CPM the blades are directly attached to the Cisco and Cisco by default has spanning tree enabled to avoid circular references. Each time the node enables the NIC, the link lights up and the Cisco starts to process spanning tree as if you plugged in a another switch. Cisco's have a setting for hosts so that this does not happen. It should be enabled for any port that connects to a host. The RH pump to get an IP via DHCP times out too quickly for Cisco with spanning tree enabled. RH should really have an option to increase the timeout. > -----Original Message----- > From: xca...@li... > [mailto:xca...@li...] On Behalf Of > Michael A Bugaev > Sent: Tuesday, June 14, 2005 5:20 AM > To: xCAT Users Mailing list > Subject: Re: [xcat-user] BladeCenter with Copper Pass-thru module > Importance: High > > > Hi Scott, > > Both chassis are connected to Cisco 6509. I'm not sure > about spanning > tree on it, it's not under my control, need to ask network > engineer. But > why it does not happen with second chassis, even thought > there is extra > switch between blade and head-node comparing to chassis with > CPM? Doesn't > it make that one more likely to suffer long delays? > > In any case, thanks for an idea. > > Regards, > Michael > > On Mon, 13 Jun 2005, Scott Denham wrote: > > > > > > > > > > > > > What's the copper pass thru plugged into? Does it have > spanning-tree > > protocol disabled? The long delays incurred by spanning > tree analyzing the > > network for loops is known to cause this timeout in the RH > installer.... > > > > > > ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- > =- =- =- =- =- > > =- > > || Scott S. Denham > > || HPC Technical Architect > > || Petroleum Lead - Deep Computing Technical Team > > || IBM Systems & Technology Group > > || Two Riverway Email: > sd...@us... > > || Houston, TX 77056 Voice/Fax: 713.940.1178 > > || > (CST / > > GMT - 6) > > ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- > =- =- =- =- =- > > =- > > > > > > > > > > Michael Bugaev > > <mabugaev@unity.n > > csu.edu> > To > > Sent by: xCAT Users Mailing list > > xcat-user-bounces <xca...@li...> > > @lists.xcat.org > cc > > > > > Subject > > 06/13/2005 03:42 [xcat-user] > BladeCenter with Copper > > PM Pass-thru module > > > > > > Please respond to > > xCAT Users > > Mailing list > > > > > > > > > > > > > > Hi all, > > > > Have any of you seen the following behavior:? > > > > BladeCenter (BC) chassis has Ethernet Switch Module (ESM) > in bay1 and > > Copper Pass-thru Module (CPM) in bay2. Blade 8843-E2U in > chassis would > > like to be loaded with RHEL3 via CPM (eth1). All necessary > settings are > > there, I believe, like "eth1" in noderes.tab. However after getting > > address from DHCPD and loading install kernel, it fails to > get address > > for eth1 second time during kickstart procedure and goes to > interactive > > install. And BTW, if I specify IP address it supposed to > get from DHCP > > by hand, I'm able to mount install directory from head-node > and continue > > interactive install. > > If I put this blade in the chassis with ESM in both bays, everything > > goes as expected. What am I missing here? > > > > Thanks in advance for any suggestions, > > Regards, > > Michael > > _______________________________________________ > > xcat-user mailing list > > xca...@li... > > http://www.xcat.org/mailman/listinfo/xcat-user > > > _______________________________________________ > xcat-user mailing list > xca...@li... > http://www.xcat.org/mailman/listinfo/xcat-user > |
|
From: mabugaev at unity.ncsu.e. (M. Bugaev) - 2005-06-14 08:42:44
|
Aha! Thanks, Egan, for clearing this up! I'll talk to our network engineer about fixing it... Appreciate your help, Regards, Michael Egan Ford wrote: > The links between the two switches are already established when blades are > in an ESM connected to a Cisco. > > With CPM the blades are directly attached to the Cisco and Cisco by default > has spanning tree enabled to avoid circular references. Each time the node > enables the NIC, the link lights up and the Cisco starts to process spanning > tree as if you plugged in a another switch. Cisco's have a setting for > hosts so that this does not happen. It should be enabled for any port that > connects to a host. > > The RH pump to get an IP via DHCP times out too quickly for Cisco with > spanning tree enabled. > > RH should really have an option to increase the timeout. > > >>-----Original Message----- >>From: xca...@li... >>[mailto:xca...@li...] On Behalf Of >>Michael A Bugaev >>Sent: Tuesday, June 14, 2005 5:20 AM >>To: xCAT Users Mailing list >>Subject: Re: [xcat-user] BladeCenter with Copper Pass-thru module >>Importance: High >> >> >>Hi Scott, >> >> Both chassis are connected to Cisco 6509. I'm not sure >>about spanning >>tree on it, it's not under my control, need to ask network >>engineer. But >>why it does not happen with second chassis, even thought >>there is extra >>switch between blade and head-node comparing to chassis with >>CPM? Doesn't >>it make that one more likely to suffer long delays? >> >>In any case, thanks for an idea. >> >>Regards, >>Michael >> >>On Mon, 13 Jun 2005, Scott Denham wrote: >> >> >>> >>> >>> >>> >>>What's the copper pass thru plugged into? Does it have >> >>spanning-tree >> >>>protocol disabled? The long delays incurred by spanning >> >>tree analyzing the >> >>>network for loops is known to cause this timeout in the RH >> >>installer.... >> >>> >>>||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- >> >>=- =- =- =- =- >> >>>=- >>>|| Scott S. Denham >>>|| HPC Technical Architect >>>|| Petroleum Lead - Deep Computing Technical Team >>>|| IBM Systems & Technology Group >>>|| Two Riverway Email: >> >>sd...@us... >> >>>|| Houston, TX 77056 Voice/Fax: 713.940.1178 >>>|| >> >> (CST / >> >>>GMT - 6) >>>||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- >> >>=- =- =- =- =- >> >>>=- >>> >>> >>> >>> >>> Michael Bugaev >>> <mabugaev@unity.n >>> csu.edu> >> >> To >> >>> Sent by: xCAT Users Mailing list >>> xcat-user-bounces <xca...@li...> >>> @lists.xcat.org >> >> cc >> >>> >> >> Subject >> >>> 06/13/2005 03:42 [xcat-user] >> >>BladeCenter with Copper >> >>> PM Pass-thru module >>> >>> >>> Please respond to >>> xCAT Users >>> Mailing list >>> >>> >>> >>> >>> >>> >>>Hi all, >>> >>> Have any of you seen the following behavior:? >>> >>>BladeCenter (BC) chassis has Ethernet Switch Module (ESM) >> >>in bay1 and >> >>>Copper Pass-thru Module (CPM) in bay2. Blade 8843-E2U in >> >>chassis would >> >>>like to be loaded with RHEL3 via CPM (eth1). All necessary >> >>settings are >> >>>there, I believe, like "eth1" in noderes.tab. However after getting >>>address from DHCPD and loading install kernel, it fails to >> >>get address >> >>>for eth1 second time during kickstart procedure and goes to >> >>interactive >> >>>install. And BTW, if I specify IP address it supposed to >> >>get from DHCP >> >>>by hand, I'm able to mount install directory from head-node >> >>and continue >> >>>interactive install. >>>If I put this blade in the chassis with ESM in both bays, everything >>>goes as expected. What am I missing here? >>> >>>Thanks in advance for any suggestions, >>>Regards, >>>Michael >>>_______________________________________________ >>>xcat-user mailing list >>>xca...@li... >>>http://www.xcat.org/mailman/listinfo/xcat-user >>> >> >>_______________________________________________ >>xcat-user mailing list >>xca...@li... >>http://www.xcat.org/mailman/listinfo/xcat-user >> > > > > _______________________________________________ > xcat-user mailing list > xca...@li... > http://www.xcat.org/mailman/listinfo/xcat-user > |
|
From: bfulda at 3drc.c. (B. Fulda) - 2005-06-13 08:03:48
|
Yep, that's what I have in my conserver.cf file. Still hangs on "waiting on telnet connection" This only started happening when I wanted a different install than what it was doing by default. It apparently was giving me the server install and I need the "everything" install. When I made the changes, added @ Everything to the to my /install/ ... compute file, that's when it wouldn't finish a sw load and hangs at this telnet msg. I'll let you know as the day progresses. Tia Brad ________________________________ From: xca...@li... [mailto:xca...@li...] On Behalf Of Egan Ford Sent: Sunday, June 12, 2005 11:59 PM To: 'xCAT Users Mailing list' Subject: RE: [xcat-user] Software loads Did you use: nodename:|conserver.telnet nodename::&: in $XCATROOT/etc/conserver.cf? That is the only reason I can think of that would have RH9 install prompting for telnet. This is used for remote console access to machines that do not have terminal servers or SOL, but that you want to still monitor the install. -----Original Message----- From: xca...@li... [mailto:xca...@li...] On Behalf Of Brad Fulda Sent: Friday, June 10, 2005 10:36 AM To: Scott Denham; xCAT Users Mailing list Subject: RE: [xcat-user] Software loads Importance: High You might be right. After I inserted "@Everything" into /install/rh9/x86/compute file. I get the blue screen and "waiting on telnet connection" and it stays there waiting. RH9 has the ability to load "Everything" I read. SUSE doesn't. Maybe I'll try deleting the other @ stuff and just leave the @Everything. At least before I was getting a load, now I'm hanging. Tia for any help Brad ________________________________ From: Scott Denham [mailto:sd...@us...] Sent: Friday, June 10, 2005 11:21 AM To: xCAT Users Mailing list Cc: Brad Fulda Subject: Re: [xcat-user] Software loads Brad - This file is documented in the Red Hat System Administration Guide. I know the installer has a checkbox for "Everything", but I don't find a comparable item in the comps.xml file. Maybe you just need to grep all the groups out of there and insert them. In doing so, I'll bet you find some stuff you don't *really* want on your cluster nodes. Yes, disk is cheap, but the more stuff you install, the longer it takes :) Here's a helpful extract regarding package selection: Packages can be specified by group or by individual package name. The installation program defines several groups that contain related packages. Refer to the RedHat/base/comps.xml file on the first Red Hat Enterprise Linux CD-ROM for a list of groups. Each group has an id, user visibility value, name, description, and package list. In the package list, the packages marked as mandatory are always installed if the group is selected, the packages marked default are selected by default if the group is selected, and the packages marked optional must be specifically selected even if the group is selected to be installed. ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- || Scott S. Denham || HPC Technical Architect || Petroleum Lead - Deep Computing Technical Team || IBM Systems & Technology Group || Two Riverway Email: sd...@us... || Houston, TX 77056 Voice/Fax: 713.940.1178 || (CST / GMT - 6) ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- xca...@li... wrote on 06/10/2005 10:50:24 AM: > Hey all, > > I have now a functioning xcat server. I still have a few bugs to > work out, like wcon doesn't work and so on, but one thing I need to > fix is the SW that is installed. When a node is installed I only > get cli. I need everything. All three disks of RH9. Where do I > make this change? I see in the /install directory there is a file on > down that sets up things like type of install, how to set up the > disks, etc. But what is the nomenclature for an "ALL" SW load. Tia > Brad Fulda > 256-651-5949_______________________________________________ > xcat-user mailing list > xca...@li... > http://www.xcat.org/mailman/listinfo/xcat-user -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.xcat.org/pipermail/xcat-user/attachments/20050613/29dda51b/attachment-0001.htm |
|
From: egan at sense.n. (E. Ford) - 2005-06-21 08:30:23
|
Perhaps I missed it, what the the status of this? Thanks. -----Original Message----- From: xca...@li... [mailto:xca...@li...] On Behalf Of Brad Fulda Sent: Monday, June 13, 2005 8:04 AM To: xCAT Users Mailing list Subject: RE: [xcat-user] Software loads Importance: High Yep, that's what I have in my conserver.cf file. Still hangs on "waiting on telnet connection" This only started happening when I wanted a different install than what it was doing by default. It apparently was giving me the server install and I need the "everything" install. When I made the changes, added @ Everything to the to my /install/ . compute file, that's when it wouldn't finish a sw load and hangs at this telnet msg. I'll let you know as the day progresses. Tia Brad _____ From: xca...@li... [mailto:xca...@li...] On Behalf Of Egan Ford Sent: Sunday, June 12, 2005 11:59 PM To: 'xCAT Users Mailing list' Subject: RE: [xcat-user] Software loads Did you use: nodename:|conserver.telnet nodename::&: in $XCATROOT/etc/conserver.cf? That is the only reason I can think of that would have RH9 install prompting for telnet. This is used for remote console access to machines that do not have terminal servers or SOL, but that you want to still monitor the install. -----Original Message----- From: xca...@li... [mailto:xca...@li...] On Behalf Of Brad Fulda Sent: Friday, June 10, 2005 10:36 AM To: Scott Denham; xCAT Users Mailing list Subject: RE: [xcat-user] Software loads Importance: High You might be right. After I inserted "@Everything" into /install/rh9/x86/compute file. I get the blue screen and "waiting on telnet connection" and it stays there waiting. RH9 has the ability to load "Everything" I read. SUSE doesn't. Maybe I'll try deleting the other @ stuff and just leave the @Everything. At least before I was getting a load, now I'm hanging. Tia for any help Brad _____ From: Scott Denham [mailto:sd...@us...] Sent: Friday, June 10, 2005 11:21 AM To: xCAT Users Mailing list Cc: Brad Fulda Subject: Re: [xcat-user] Software loads Brad - This file is documented in the Red Hat System Administration Guide. I know the installer has a checkbox for "Everything", but I don't find a comparable item in the comps.xml file. Maybe you just need to grep all the groups out of there and insert them. In doing so, I'll bet you find some stuff you don't *really* want on your cluster nodes. Yes, disk is cheap, but the more stuff you install, the longer it takes :) Here's a helpful extract regarding package selection: Packages can be specified by group or by individual package name. The installation program defines several groups that contain related packages. Refer to the RedHat/base/comps.xml file on the first Red Hat Enterprise Linux CD-ROM for a list of groups. Each group has an id, user visibility value, name, description, and package list. In the package list, the packages marked as mandatory are always installed if the group is selected, the packages marked default are selected by default if the group is selected, and the packages marked optional must be specifically selected even if the group is selected to be installed. ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- || Scott S. Denham || HPC Technical Architect || Petroleum Lead - Deep Computing Technical Team || IBM Systems & Technology Group || Two Riverway Email: sd...@us... || Houston, TX 77056 Voice/Fax: 713.940.1178 || (CST / GMT - 6) ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- xca...@li... wrote on 06/10/2005 10:50:24 AM: > Hey all, > > I have now a functioning xcat server. I still have a few bugs to > work out, like wcon doesn't work and so on, but one thing I need to > fix is the SW that is installed. When a node is installed I only > get cli. I need everything. All three disks of RH9. Where do I > make this change? I see in the /install directory there is a file on > down that sets up things like type of install, how to set up the > disks, etc. But what is the nomenclature for an "ALL" SW load. Tia > Brad Fulda > 256-651-5949_______________________________________________ > xcat-user mailing list > xca...@li... > http://www.xcat.org/mailman/listinfo/xcat-user -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.xcat.org/pipermail/xcat-user/attachments/20050621/768f88ed/attachment.htm |
|
From: bfulda at 3drc.c. (B. Fulda) - 2005-06-21 08:52:27
|
Hey Egan, It is working now. Some tab fies were modified, namely the nodelist and conserver.cf files. Kevin Bryant came down and we worked Fulda hours(18+) on the server and finally got it working and the Blades(144) installed. Kevin knew xcat very well and it was a great benefit to have him here. I will be writing a blade only installation version of xcat and hopefully you will review it for me. Tia Brad ________________________________ From: xca...@li... [mailto:xca...@li...] On Behalf Of Egan Ford Sent: Tuesday, June 21, 2005 9:30 AM To: 'xCAT Users Mailing list' Subject: RE: [xcat-user] Software loads Perhaps I missed it, what the the status of this? Thanks. -----Original Message----- From: xca...@li... [mailto:xca...@li...] On Behalf Of Brad Fulda Sent: Monday, June 13, 2005 8:04 AM To: xCAT Users Mailing list Subject: RE: [xcat-user] Software loads Importance: High Yep, that's what I have in my conserver.cf file. Still hangs on "waiting on telnet connection" This only started happening when I wanted a different install than what it was doing by default. It apparently was giving me the server install and I need the "everything" install. When I made the changes, added @ Everything to the to my /install/ ... compute file, that's when it wouldn't finish a sw load and hangs at this telnet msg. I'll let you know as the day progresses. Tia Brad ________________________________ From: xca...@li... [mailto:xca...@li...] On Behalf Of Egan Ford Sent: Sunday, June 12, 2005 11:59 PM To: 'xCAT Users Mailing list' Subject: RE: [xcat-user] Software loads Did you use: nodename:|conserver.telnet nodename::&: in $XCATROOT/etc/conserver.cf? That is the only reason I can think of that would have RH9 install prompting for telnet. This is used for remote console access to machines that do not have terminal servers or SOL, but that you want to still monitor the install. -----Original Message----- From: xca...@li... [mailto:xca...@li...] On Behalf Of Brad Fulda Sent: Friday, June 10, 2005 10:36 AM To: Scott Denham; xCAT Users Mailing list Subject: RE: [xcat-user] Software loads Importance: High You might be right. After I inserted "@Everything" into /install/rh9/x86/compute file. I get the blue screen and "waiting on telnet connection" and it stays there waiting. RH9 has the ability to load "Everything" I read. SUSE doesn't. Maybe I'll try deleting the other @ stuff and just leave the @Everything. At least before I was getting a load, now I'm hanging. Tia for any help Brad ________________________________ From: Scott Denham [mailto:sd...@us...] Sent: Friday, June 10, 2005 11:21 AM To: xCAT Users Mailing list Cc: Brad Fulda Subject: Re: [xcat-user] Software loads Brad - This file is documented in the Red Hat System Administration Guide. I know the installer has a checkbox for "Everything", but I don't find a comparable item in the comps.xml file. Maybe you just need to grep all the groups out of there and insert them. In doing so, I'll bet you find some stuff you don't *really* want on your cluster nodes. Yes, disk is cheap, but the more stuff you install, the longer it takes :) Here's a helpful extract regarding package selection: Packages can be specified by group or by individual package name. The installation program defines several groups that contain related packages. Refer to the RedHat/base/comps.xml file on the first Red Hat Enterprise Linux CD-ROM for a list of groups. Each group has an id, user visibility value, name, description, and package list. In the package list, the packages marked as mandatory are always installed if the group is selected, the packages marked default are selected by default if the group is selected, and the packages marked optional must be specifically selected even if the group is selected to be installed. ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- || Scott S. Denham || HPC Technical Architect || Petroleum Lead - Deep Computing Technical Team || IBM Systems & Technology Group || Two Riverway Email: sd...@us... || Houston, TX 77056 Voice/Fax: 713.940.1178 || (CST / GMT - 6) ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- xca...@li... wrote on 06/10/2005 10:50:24 AM: > Hey all, > > I have now a functioning xcat server. I still have a few bugs to > work out, like wcon doesn't work and so on, but one thing I need to > fix is the SW that is installed. When a node is installed I only > get cli. I need everything. All three disks of RH9. Where do I > make this change? I see in the /install directory there is a file on > down that sets up things like type of install, how to set up the > disks, etc. But what is the nomenclature for an "ALL" SW load. Tia > Brad Fulda > 256-651-5949_______________________________________________ > xcat-user mailing list > xca...@li... > http://www.xcat.org/mailman/listinfo/xcat-user -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.xcat.org/pipermail/xcat-user/attachments/20050621/f56e7e93/attachment-0001.htm |
|
From: dcumming at us.ibm.c. (D. Cummings) - 2005-06-10 10:01:40
|
Skipped content of type multipart/alternative-------------- next part -------------- A non-text attachment was scrubbed... Name: graycol.gif Type: image/gif Size: 105 bytes Desc: not available Url : http://www.xcat.org/pipermail/xcat-user/attachments/20050610/3f0e99d5/graycol.gif -------------- next part -------------- A non-text attachment was scrubbed... Name: pic30826.gif Type: image/gif Size: 1255 bytes Desc: not available Url : http://www.xcat.org/pipermail/xcat-user/attachments/20050610/3f0e99d5/pic30826.gif -------------- next part -------------- A non-text attachment was scrubbed... Name: ecblank.gif Type: image/gif Size: 45 bytes Desc: not available Url : http://www.xcat.org/pipermail/xcat-user/attachments/20050610/3f0e99d5/ecblank.gif |
|
From: sdenham at us.ibm.c. (S. Denham) - 2005-06-10 10:21:19
|
Brad - This file is documented in the Red Hat System Administration Guide. I know the installer has a checkbox for "Everything", but I don't find a comparable item in the comps.xml file. Maybe you just need to grep all the groups out of there and insert them. In doing so, I'll bet you find some stuff you don't *really* want on your cluster nodes. Yes, disk is cheap, but the more stuff you install, the longer it takes :) Here's a helpful extract regarding package selection: Packages can be specified by group or by individual package name. The installation program defines several groups that contain related packages. Refer to the RedHat/base/comps.xml file on the first Red Hat Enterprise Linux CD-ROM for a list of groups. Each group has an id, user visibility value, name, description, and package list. In the package list, the packages marked as mandatory are always installed if the group is selected, the packages marked default are selected by default if the group is selected, and the packages marked optional must be specifically selected even if the group is selected to be installed. ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- || Scott S. Denham || HPC Technical Architect || Petroleum Lead - Deep Computing Technical Team || IBM Systems & Technology Group || Two Riverway Email: sd...@us... || Houston, TX 77056 Voice/Fax: 713.940.1178 || (CST / GMT - 6) ||=- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- =- xca...@li... wrote on 06/10/2005 10:50:24 AM: > Hey all, > > I have now a functioning xcat server. I still have a few bugs to > work out, like wcon doesn?t work and so on, but one thing I need to > fix is the SW that is installed. When a node is installed I only > get cli. I need everything. All three disks of RH9. Where do I > make this change? I see in the /install directory there is a file on > down that sets up things like type of install, how to set up the > disks, etc. But what is the nomenclature for an ?ALL? SW load. Tia > Brad Fulda > 256-651-5949_______________________________________________ > xcat-user mailing list > xca...@li... > http://www.xcat.org/mailman/listinfo/xcat-user -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.xcat.org/pipermail/xcat-user/attachments/20050610/d7b3042d/attachment.htm |
|
From: jburk at mainframe.ca (J. Burk) - 2005-06-10 12:01:37
|
you'll need to change the last field in your nodetype.tab, the one that points to the image. the format is nodeName osver,arch,image you probably have something like: r01c01b01 rh9,x86,compute the images are *.tmpl files, and live in /opt/xcat/install/<osver>/<arch>/base. There is an 'all.tmpl' file, so change the last field in your nodelist file from 'compute' to 'all', re-run 'nodeset <node> install' and 'rreset <node>', and you should be good to go. If you ever need to customize or create a new .tmpl file, copy it from /opt/xcat/install/<osver>/<arch>/base up to /opt/xcat/install/<osver>/<arch>. Template files in the parent directory will override the ones in base/ jburk. Brad Fulda wrote: > Hey all, > > I have now a functioning xcat server. I still have a few bugs to work > out, like wcon doesn?t work and so on, but one thing I need to fix is > the SW that is installed. When a node is installed I only get cli. I > need everything. All three disks of RH9. Where do I make this change? > I see in the /install directory there is a file on down that sets up > things like type of install, how to set up the disks, etc. But what is > the nomenclature for an ?ALL? SW load. Tia > > Brad Fulda > > 256-651-5949 > >------------------------------------------------------------------------ > >_______________________________________________ >xcat-user mailing list >xca...@li... >http://www.xcat.org/mailman/listinfo/xcat-user > > |
Thanks for helping keep SourceForge clean.
X