Hi Jimi:
OSCAR 4.2 does not support
Fedora Core 4, if you want Fedora Core 4 support, please use the 4.2.1 nightly
tarball:
http://oscar.openclustergroup.org/filebrowser/49/branch
http://oscar.openclustergroup.org/filebrowser/49/branch
If you're still having issues
with 4.2.1 and Fedora Core 4, please post your /opt/oscar/oscarinstall.log
and/or installation typescript so that we can debug your issue:
http://oscar.openclustergroup.org/faq.installation.typescript
http://oscar.openclustergroup.org/faq.installation.typescript
(You might want to compress
the file before sending it)
BTW, please make sure that
when you are installing the OSes, that you choose the "workstation" installation
where available - this will provide the base OS installation for OSCAR to work
with.
OSCAR 5.0 is not released
yet. However, if you would like to out our *development* code, you are
welcome to download the nightly tarballs available here:
http://oscar.openclustergroup.org/filebrowser/49/trunk
http://oscar.openclustergroup.org/filebrowser/49/trunk
(Please read the README
before you download anything...)
Please note that OSCAR 5.0 is
_very different_ from previous versions. The documentation right now is
quite lacking, but some information is available via our Trac site http://svn.oscar.openclustergroup.org/trac/oscar.
You are always welcome to ask us questions here or on our IRC channel
#oscar-cluster @irc.freenode.net.
Currently we have an issue
with Fedora Core 4 and trunk, so if you want to try trunk out, I recommend
Fedora Core 5.
Lastly, Erich Focht has
created a VMware appliance for OSCAR headnode, you might want to give it a try
too:
http://oscar.openclustergroup.org/news.oscar-appliance
http://oscar.openclustergroup.org/news.oscar-appliance
Happy OSCARing.
Cheers,
Bernard
From: [EMAIL PROTECTED] on behalf of Jimi Fayemi
Sent: Tue 20/06/2006 21:38
To: oscar-users@lists.sourceforge.net
Subject: [Oscar-users] Installation Script failure
Hi,
I
am kind of a newbie and have found it difficult to install OSCAR 4.2, it keeps
failing during the install_cluster script installation. Anyway I have tried FC3
and FC4, it fails to install Perl-DBD-MySQL, I have checked and see that the
package is available at /tftpboot/rpm, any advice for me? I was also inform
that OSCAR 5.0 is available, any idea where I can download it.
Thanks,
'Jimi
-----
Original Message ----
From: [EMAIL PROTECTED]
To: oscar-users@lists.sourceforge.net
Sent: Tuesday, June 20, 2006 3:02:36 PM
Subject: Oscar-users Digest, Vol 1, Issue 1694
From: [EMAIL PROTECTED]
To: oscar-users@lists.sourceforge.net
Sent: Tuesday, June 20, 2006 3:02:36 PM
Subject: Oscar-users Digest, Vol 1, Issue 1694
Send Oscar-users mailing list submissions
to
oscar-users@lists.sourceforge.net
To subscribe or unsubscribe via the World Wide Web, visit
https://lists.sourceforge.net/lists/listinfo/oscar-users
or, via email, send a message with subject or body 'help' to
[EMAIL PROTECTED]
You can reach the person managing the list at
[EMAIL PROTECTED]
When replying, please edit your Subject line so it is more specific
than "Re: Contents of Oscar-users digest..."
Today's Topics:
1. How to update a head node and nodes (Brad Aisa)
2. Re: Network Boot Problem & Grub (Tyler Cruickshank)
3. Re: [Oscar-devel] How to keep head/nodes up to date? (Erich Focht)
----------------------------------------------------------------------
Message: 1
Date: Mon, 19 Jun 2006 16:46:03 -0700 (PDT)
From: Brad Aisa <[EMAIL PROTECTED]>
Subject: [Oscar-users] How to update a head node and nodes
To: oscar users <oscar-users@lists.sourceforge.net>
Message-ID: <[EMAIL PROTECTED]>
Content-Type: text/plain; charset=us-ascii
[asking in user and devel since this pertains to Oscar 5, but may be a generic question that is as applicable to the current release version of Oscar]
Sorry I keep asking this, hope I'm not trying folks' patience...
But I am still very clueless on updating my headnode and my nodes. I did a default installation based on the base level Fedora distribution, but now I want to update everything.
I am afraid to do anything, for fear of completely messing up my cluster, but at the same time, I am really quite urgently in need of many updates.
Thanks for any tips!
BTW, I really think this topic is urgently needed in the Oscar documentation. If someone can explain the principles to me, I would be happy to write up a detailed explanation suitable for the documentation.
Brad Aisa
baisa at brad-aisa dot com
------------------------------
Message: 2
Date: Tue, 20 Jun 2006 08:12:28 -0600
From: "Tyler Cruickshank" <[EMAIL PROTECTED]>
Subject: Re: [Oscar-users] Network Boot Problem & Grub
To: "Bernard Li" <[EMAIL PROTECTED]>,<oscar-users@lists.sourceforge.net>
Message-ID: <[EMAIL PROTECTED]>
Content-Type: text/plain; charset="iso-8859-1"
Bernard,
Yes, both SCSI and IDE drives. Both drives are the same size so it makes no difference to me which it boots from.
ty
>>> "Bernard Li" <[EMAIL PROTECTED]> 6/19/2006 5:27 PM >>>
Hi Tyler:
So your compute nodes have _both_ SCSI and IDE drives? Which drive (SCSI or IDE) do you want your compute nodes to boot off?
Cheers,
Bernard
From: Tyler Cruickshank [mailto:[EMAIL PROTECTED]
Sent: Monday, June 19, 2006 15:01
To: Bernard Li; oscar-users@lists.sourceforge.net
Subject: RE: [Oscar-users] Network Boot Problem & Grub
Thank you Bernard. I found menu.lst. It says that the kernel is on the scsi drive but the scsi drive does not seem to be showing up on the node's F10 device configuration menu. So, we disconnected the scsi drive and PXE booted again. This time, after a reboot, we were successful! For whatever reason, the kernel was placed on the scsi, but this was not recognized upon boot. This will be fine as is, but perhaps there is an easy fix to the scsi problem? The server node also has a scsi and ide drive, but it is a much newer machine compared to the ~3-4 year old client node. Maybe this has nothing to do with it though?
-ty
menu.lst File:
/mnt/sysimage/boot/grub/menu.lst
*----------------------------------------------------------------
# Splash image
splashimage=(hd1,0)/grub/splash.xpm.gz
# kernel0
title 2.6.9-34.ELsmp_(hd1,0)
root (hd1,0)
kernel /vmlinuz-2.6.9-34.ELsmp ro root=/dev/sda6
initrd /sc-initrd-2.6.9-34.ELsmp.gz
# kernel1
title 2.6.9-34.EL_(hd1,0)
root (hd1,0)
kernel /vmlinuz-2.6.9-34.EL ro root=/dev/sda6
initrd /sc-initrd-2.6.9-34.EL.gz
*----------------------------------------------------------------
>>> "Bernard Li" <[EMAIL PROTECTED]> 6/19/2006 2:49:33 PM >>>
Hi Tyler:
/boot should be created, however, it's possible that it was simply not mounted.
When you boot into rescue mode, it'll ask whether you want it to search for installed Red Hat OS on the HD, if you say yes it should automatically mount all the partitions under /mnt/sysimage (or something like that). So you should look inside /mnt/sysimage/boot in that case.
If it did not, you can manually mount it by doing:
mkdir -p /mnt/boot
mount /dev/hda1 /mnt/boot (substitute "sda" with "hda" if you're using SCSI HD).
Then cd into /mnt/boot and look for grub/menu.lst, it should be there...
As for the directory listing you showed from the "ash shell", the "boot" directory should reside in /a/boot if it was mounted. /a/ is the temporary mount point for your compute node's /root partition.
Cheers,
Bernard
From: Tyler Cruickshank [mailto:[EMAIL PROTECTED]
Sent: Monday, June 19, 2006 13:12
To: Bernard Li; oscar-users@lists.sourceforge.net
Subject: RE: [Oscar-users] Network Boot Problem & Grub
Bernard,
Thank you again for your speedy and helpful responses. We all appreciate your efforts.
Well, no wonder I couldn't find the .img ... I dont have a /boot directory. I went ahead and used cd 1 to boot into rescue mode. No menu.lst file was created and no /boot directory exists. I can show you what directories were created ... from the ash shell:
a/ etc/ linuxrc* old_root/ sbin/ usr/
bin/ floppy/ my_modules/ proc/ scripts/ var/
dev/ lib/ new_root/ root/ tmp/
Thoughts on this dilemma?
Thanks.
-ty
>>> "Bernard Li" <[EMAIL PROTECTED]> 6/19/2006 1:50:50 PM >>>
Hi Tyler:
Do you have both IDE and SCSI HDs in your compute nodes?
Can you boot one of your compute nodes (which failed installation) into rescue mode (using the first CD of RHEL4 and typing "linux rescue" at the prompt) and show us the contents of /boot/grub/menu.lst (of the HD).
It looks like Grub was not installed properly...
P.S. Kernel and initrd.img of your installed node should reside in /boot - SystemConfigurator creates a ramdisk for you on the fly though, and the filename is prefixed by "sc-".
Cheers,
Bernard
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Tyler Cruickshank
Sent: Monday, June 19, 2006 8:01
To: oscar-users@lists.sourceforge.net
Subject: [Oscar-users] Network Boot Problem & Grub
Hi.
I have successfully created an image and am now trying to network boot a single node. The network install finishes (apparently with success) but it does give some "problem" messages before the boot process is done (see below). After the network boot, I reboot again (from the hard drive), but I get a grub> prompt. From the prompt, I have tried to load the kernel but I dont know where it lives. I have also tried to "find" the image, based on the name I originally gave it, but it cant seem to find it. So, it seems something is not right. Where should I look to investigate the problem further?
Thanks.
-Tyler
Utah Division of Air Quality
*-------------------------------------------------------------------------------------------
mv: cannot stat '/boot/grub/device.map' : No such file or directory
Probing deivices to guess BIOS drives. This may take a long time.
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
Installation finished. No error reported.
This is the contents of the device map /boot/grub/device.map.
.. blah blah blah ...
(fd0) /dev/flopy/0
(hd0) /dev/ide/host0/bus0/target0/lun0/disc
(hd1) /dev/scsi/host0/bus0/target0/lun0/disc
Use of unitintialized value in concatenation (.) or string at /usr/lib/systemcnfig/Boot/Grub.pm line 315
Probing deivices to guess BIOS drives. This may take a long time.
end_request: I/O error, dev fdo, sector 0
run_post_install_scripts
... blah blah ...
Reboot me already.
*-------------------------------------------------------------------------------------------
Finally, is there a file created that contains all of the network boot screen output? I found one file in /tmp called si.log, but it contains only a small bit of the process.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://sourceforge.net/mailarchive/forum.php?forum=oscar-users/attachments/20060620/3e92eba3/attachment.html
------------------------------
Message: 3
Date: Tue, 20 Jun 2006 10:50:51 +0200
From: Erich Focht <[EMAIL PROTECTED]>
Subject: Re: [Oscar-users] [Oscar-devel] How to keep head/nodes up to
date?
To: oscar-devel@lists.sourceforge.net
Cc: Brad Aisa <[EMAIL PROTECTED]>, oscar users
<oscar-users@lists.sourceforge.net>
Message-ID: <[EMAIL PROTECTED]>
Content-Type: text/plain; charset="iso-8859-1"
Hi Brad,
On Sunday 18 June 2006 07:50, Brad Aisa wrote:
> [posted also to devel because I'm running FC5/Oscar 5]
>
> I've discovered my first big issue today with my ancient FC5 kernel (2054)
> (can't compile the nvidia driver, known bug with this kernel version) -- I
> want to update my system, or at least parts of it. I'll need other stuff
> updated too I know for sure (gcc, other stuff).
>
> But there were dire warnings about not updating a system when using OSCAR. I
> could find no section in the Oscar manuals on the overall approach one needs
> to take.
Sorry for the missing documentation, we'll try to improve that ASAP. The
warnings about updating an OSCAR system should not be taken too
seriously. Updating is possible but with OSCAR versions <5 you should better
know what you're doing.
The main reason for having warnings here is that you get quickly into trouble
if your repository packages are newer than those installed on the master
node. And vice-versa.
> Can I update my system using yum to the latest version of FC5???
Yes. But after updating the master node you MUST update your distro repository
(/tftpboot/distro/fedora-5-i386/). In case you use remote repositories
(/tftpboot/distro/fedora-5-i386.url file) you must add to it the URL of a
FC5 updates repository.
If you use normal (local) repositories (where you copied your RPMs from the
DVD to /tftpboot/distro/fedora-5-i386/) $OSCAR_HOME/scripts/repo-update makes
your life easier when updating. What you need to do is:
cd /tftpboot/distro/fedora-5-i386/
repo-update --url http://mirrors.dotsrc.org/fedora/updates/5/i386/ --rmdup
> Would it be better if I do this now, and then rebuild my kernel images?
> (Before I do my actual cluster install?)
Yas. That simplifies things a lot!
> How do I make sure the new rpms are the ones in the image? (I copied the DVD
> to the tftpboot folder, but if I do yum, then I won't know where the
> packages are)
If you use "yume" instead of "yum" you will be using the repositories known by
OSCAR, i.e. those in /tftpboot/distro/*.
> Just as a comment, the installation manual is great, but this entire aspect of updating a system is important and doesn't seem to be treated anywhere. I would highly recommend adding an explanatory section to the installation manual.
Thanks for pointing this out. As said, it is possible, and we'll add
documentation on this really soon. It will go to the wiki.
Best regards,
Erich
------------------------------
------------------------------
_______________________________________________
Oscar-users mailing list
Oscar-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oscar-users
End of Oscar-users Digest, Vol 1, Issue 1694
********************************************
oscar-users@lists.sourceforge.net
To subscribe or unsubscribe via the World Wide Web, visit
https://lists.sourceforge.net/lists/listinfo/oscar-users
or, via email, send a message with subject or body 'help' to
[EMAIL PROTECTED]
You can reach the person managing the list at
[EMAIL PROTECTED]
When replying, please edit your Subject line so it is more specific
than "Re: Contents of Oscar-users digest..."
Today's Topics:
1. How to update a head node and nodes (Brad Aisa)
2. Re: Network Boot Problem & Grub (Tyler Cruickshank)
3. Re: [Oscar-devel] How to keep head/nodes up to date? (Erich Focht)
----------------------------------------------------------------------
Message: 1
Date: Mon, 19 Jun 2006 16:46:03 -0700 (PDT)
From: Brad Aisa <[EMAIL PROTECTED]>
Subject: [Oscar-users] How to update a head node and nodes
To: oscar users <oscar-users@lists.sourceforge.net>
Message-ID: <[EMAIL PROTECTED]>
Content-Type: text/plain; charset=us-ascii
[asking in user and devel since this pertains to Oscar 5, but may be a generic question that is as applicable to the current release version of Oscar]
Sorry I keep asking this, hope I'm not trying folks' patience...
But I am still very clueless on updating my headnode and my nodes. I did a default installation based on the base level Fedora distribution, but now I want to update everything.
I am afraid to do anything, for fear of completely messing up my cluster, but at the same time, I am really quite urgently in need of many updates.
Thanks for any tips!
BTW, I really think this topic is urgently needed in the Oscar documentation. If someone can explain the principles to me, I would be happy to write up a detailed explanation suitable for the documentation.
Brad Aisa
baisa at brad-aisa dot com
------------------------------
Message: 2
Date: Tue, 20 Jun 2006 08:12:28 -0600
From: "Tyler Cruickshank" <[EMAIL PROTECTED]>
Subject: Re: [Oscar-users] Network Boot Problem & Grub
To: "Bernard Li" <[EMAIL PROTECTED]>,<oscar-users@lists.sourceforge.net>
Message-ID: <[EMAIL PROTECTED]>
Content-Type: text/plain; charset="iso-8859-1"
Bernard,
Yes, both SCSI and IDE drives. Both drives are the same size so it makes no difference to me which it boots from.
ty
>>> "Bernard Li" <[EMAIL PROTECTED]> 6/19/2006 5:27 PM >>>
Hi Tyler:
So your compute nodes have _both_ SCSI and IDE drives? Which drive (SCSI or IDE) do you want your compute nodes to boot off?
Cheers,
Bernard
From: Tyler Cruickshank [mailto:[EMAIL PROTECTED]
Sent: Monday, June 19, 2006 15:01
To: Bernard Li; oscar-users@lists.sourceforge.net
Subject: RE: [Oscar-users] Network Boot Problem & Grub
Thank you Bernard. I found menu.lst. It says that the kernel is on the scsi drive but the scsi drive does not seem to be showing up on the node's F10 device configuration menu. So, we disconnected the scsi drive and PXE booted again. This time, after a reboot, we were successful! For whatever reason, the kernel was placed on the scsi, but this was not recognized upon boot. This will be fine as is, but perhaps there is an easy fix to the scsi problem? The server node also has a scsi and ide drive, but it is a much newer machine compared to the ~3-4 year old client node. Maybe this has nothing to do with it though?
-ty
menu.lst File:
/mnt/sysimage/boot/grub/menu.lst
*----------------------------------------------------------------
# Splash image
splashimage=(hd1,0)/grub/splash.xpm.gz
# kernel0
title 2.6.9-34.ELsmp_(hd1,0)
root (hd1,0)
kernel /vmlinuz-2.6.9-34.ELsmp ro root=/dev/sda6
initrd /sc-initrd-2.6.9-34.ELsmp.gz
# kernel1
title 2.6.9-34.EL_(hd1,0)
root (hd1,0)
kernel /vmlinuz-2.6.9-34.EL ro root=/dev/sda6
initrd /sc-initrd-2.6.9-34.EL.gz
*----------------------------------------------------------------
>>> "Bernard Li" <[EMAIL PROTECTED]> 6/19/2006 2:49:33 PM >>>
Hi Tyler:
/boot should be created, however, it's possible that it was simply not mounted.
When you boot into rescue mode, it'll ask whether you want it to search for installed Red Hat OS on the HD, if you say yes it should automatically mount all the partitions under /mnt/sysimage (or something like that). So you should look inside /mnt/sysimage/boot in that case.
If it did not, you can manually mount it by doing:
mkdir -p /mnt/boot
mount /dev/hda1 /mnt/boot (substitute "sda" with "hda" if you're using SCSI HD).
Then cd into /mnt/boot and look for grub/menu.lst, it should be there...
As for the directory listing you showed from the "ash shell", the "boot" directory should reside in /a/boot if it was mounted. /a/ is the temporary mount point for your compute node's /root partition.
Cheers,
Bernard
From: Tyler Cruickshank [mailto:[EMAIL PROTECTED]
Sent: Monday, June 19, 2006 13:12
To: Bernard Li; oscar-users@lists.sourceforge.net
Subject: RE: [Oscar-users] Network Boot Problem & Grub
Bernard,
Thank you again for your speedy and helpful responses. We all appreciate your efforts.
Well, no wonder I couldn't find the .img ... I dont have a /boot directory. I went ahead and used cd 1 to boot into rescue mode. No menu.lst file was created and no /boot directory exists. I can show you what directories were created ... from the ash shell:
a/ etc/ linuxrc* old_root/ sbin/ usr/
bin/ floppy/ my_modules/ proc/ scripts/ var/
dev/ lib/ new_root/ root/ tmp/
Thoughts on this dilemma?
Thanks.
-ty
>>> "Bernard Li" <[EMAIL PROTECTED]> 6/19/2006 1:50:50 PM >>>
Hi Tyler:
Do you have both IDE and SCSI HDs in your compute nodes?
Can you boot one of your compute nodes (which failed installation) into rescue mode (using the first CD of RHEL4 and typing "linux rescue" at the prompt) and show us the contents of /boot/grub/menu.lst (of the HD).
It looks like Grub was not installed properly...
P.S. Kernel and initrd.img of your installed node should reside in /boot - SystemConfigurator creates a ramdisk for you on the fly though, and the filename is prefixed by "sc-".
Cheers,
Bernard
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Tyler Cruickshank
Sent: Monday, June 19, 2006 8:01
To: oscar-users@lists.sourceforge.net
Subject: [Oscar-users] Network Boot Problem & Grub
Hi.
I have successfully created an image and am now trying to network boot a single node. The network install finishes (apparently with success) but it does give some "problem" messages before the boot process is done (see below). After the network boot, I reboot again (from the hard drive), but I get a grub> prompt. From the prompt, I have tried to load the kernel but I dont know where it lives. I have also tried to "find" the image, based on the name I originally gave it, but it cant seem to find it. So, it seems something is not right. Where should I look to investigate the problem further?
Thanks.
-Tyler
Utah Division of Air Quality
*-------------------------------------------------------------------------------------------
mv: cannot stat '/boot/grub/device.map' : No such file or directory
Probing deivices to guess BIOS drives. This may take a long time.
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
end_request: I/O error, dev fd0, sector 0
Installation finished. No error reported.
This is the contents of the device map /boot/grub/device.map.
.. blah blah blah ...
(fd0) /dev/flopy/0
(hd0) /dev/ide/host0/bus0/target0/lun0/disc
(hd1) /dev/scsi/host0/bus0/target0/lun0/disc
Use of unitintialized value in concatenation (.) or string at /usr/lib/systemcnfig/Boot/Grub.pm line 315
Probing deivices to guess BIOS drives. This may take a long time.
end_request: I/O error, dev fdo, sector 0
run_post_install_scripts
... blah blah ...
Reboot me already.
*-------------------------------------------------------------------------------------------
Finally, is there a file created that contains all of the network boot screen output? I found one file in /tmp called si.log, but it contains only a small bit of the process.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://sourceforge.net/mailarchive/forum.php?forum=oscar-users/attachments/20060620/3e92eba3/attachment.html
------------------------------
Message: 3
Date: Tue, 20 Jun 2006 10:50:51 +0200
From: Erich Focht <[EMAIL PROTECTED]>
Subject: Re: [Oscar-users] [Oscar-devel] How to keep head/nodes up to
date?
To: oscar-devel@lists.sourceforge.net
Cc: Brad Aisa <[EMAIL PROTECTED]>, oscar users
<oscar-users@lists.sourceforge.net>
Message-ID: <[EMAIL PROTECTED]>
Content-Type: text/plain; charset="iso-8859-1"
Hi Brad,
On Sunday 18 June 2006 07:50, Brad Aisa wrote:
> [posted also to devel because I'm running FC5/Oscar 5]
>
> I've discovered my first big issue today with my ancient FC5 kernel (2054)
> (can't compile the nvidia driver, known bug with this kernel version) -- I
> want to update my system, or at least parts of it. I'll need other stuff
> updated too I know for sure (gcc, other stuff).
>
> But there were dire warnings about not updating a system when using OSCAR. I
> could find no section in the Oscar manuals on the overall approach one needs
> to take.
Sorry for the missing documentation, we'll try to improve that ASAP. The
warnings about updating an OSCAR system should not be taken too
seriously. Updating is possible but with OSCAR versions <5 you should better
know what you're doing.
The main reason for having warnings here is that you get quickly into trouble
if your repository packages are newer than those installed on the master
node. And vice-versa.
> Can I update my system using yum to the latest version of FC5???
Yes. But after updating the master node you MUST update your distro repository
(/tftpboot/distro/fedora-5-i386/). In case you use remote repositories
(/tftpboot/distro/fedora-5-i386.url file) you must add to it the URL of a
FC5 updates repository.
If you use normal (local) repositories (where you copied your RPMs from the
DVD to /tftpboot/distro/fedora-5-i386/) $OSCAR_HOME/scripts/repo-update makes
your life easier when updating. What you need to do is:
cd /tftpboot/distro/fedora-5-i386/
repo-update --url http://mirrors.dotsrc.org/fedora/updates/5/i386/ --rmdup
> Would it be better if I do this now, and then rebuild my kernel images?
> (Before I do my actual cluster install?)
Yas. That simplifies things a lot!
> How do I make sure the new rpms are the ones in the image? (I copied the DVD
> to the tftpboot folder, but if I do yum, then I won't know where the
> packages are)
If you use "yume" instead of "yum" you will be using the repositories known by
OSCAR, i.e. those in /tftpboot/distro/*.
> Just as a comment, the installation manual is great, but this entire aspect of updating a system is important and doesn't seem to be treated anywhere. I would highly recommend adding an explanatory section to the installation manual.
Thanks for pointing this out. As said, it is possible, and we'll add
documentation on this really soon. It will go to the wiki.
Best regards,
Erich
------------------------------
------------------------------
_______________________________________________
Oscar-users mailing list
Oscar-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oscar-users
End of Oscar-users Digest, Vol 1, Issue 1694
********************************************
_______________________________________________ Oscar-users mailing list Oscar-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oscar-users
