On Friday 20 April 2007 02:09:40 pm Bernard Li wrote: > Hi Ivan: > > On 4/20/07, Ivan Adzhubey <[EMAIL PROTECTED]> wrote: > > Thanks for help, I am catching on systemimager slowly! > > No problems. > > > Yep, I missed the fact that monitord (and netbootmond) are not registered > > as system services with chkconfig and hence not restarted automatically > > upon reboot. Looks like you are a Debian guy ;-) > > I'm not really a Debian guy, although I do run a Debian server... I > have been supporting OSCAR on Fedora, RHEL/CentOS and openSUSE. > Geoffroy is the Debian guy :-) > > Those services should be registered with chkconfig -- are you saying > that they are no longer registered after the upgrade? If so, that > needs to be fixed.
Looks like rpm update scripts removed both services for some reason. I checked the init.d scripts and they both have proper chkconfig-complaint headers, so I just added services with chkconfig --add. Worked like a charm. You might want to look into rpm scripts to figure out what's wrong with them. > However, chkconfig only comes into play when you reboot the machine. > So if you upgrade the RPMs on the spot, and the spec file does not > tell the system to restart the services, then you need to restart them > manually... I did reboot the box in the meantime that's why I thought everything should be automagically in order... > > Now I am having another problem, everything goes well until the > > provisioned node gets to reading local.cfg file, which it now tries to > > get from a floppy drive (and nodes do not have floppies at all). I guess > > I need to configure netbootmgr manually since OSCAR wizard seems to fail > > to do this after systemimager upgrade. > > I'm not sure what you mean -- can you please elaborate? Upon client's network boot, at the stage after both the kernel and initrd loaded and kernel booted successfully, a bunch of scripts are executed, and the last in sequence (I believe) is the one called read_cfg_local (I think). At this point it tries to read local config file from a floppy drive (/dev/fd0) instead of the root of HDD and fails (there is no floppy drive in a box). The provisioning process then just hangs retrying to read from fd0 and failing with I/O error messages. I don't know if it may timeout eventually, I just waited for 5 min before rebooting the node. Anyway, this behavior is not normal and should be fixed. Now I am trying to figure out how. Strangely, it did not try to do this before, I have successfully provisioned this node after systemimager update already. May be I have screwed something while editing .master script. But I don't see any errors reported... Anyway, have to look into this, will post update later. Thanks again! --Ivan > netbootmgr should work provided that systemimager-server-netbootmond > is running -- can you check to see if it is? > > > Great, that seems like a reasonable solution. I can see node<N>.sh > > scripts created in /var/lib/systemimager/scripts/ as symlinks to the > > master script in the same directory. Is that what is used to provision > > each node? I guess if I make a copy of the master script under node's > > name instead of a symlink and edit it then that's what will be used to > > provision this node, correct? If so it would be more than enough for me > > at the moment. > > Just backup the original .master script, modify it and all the nodes > will use the updated script to provision the nodes -- that's what I > would do. ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Oscar-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/oscar-devel
