From: Iain Barker (ibarker@aastra.com)
Date: Wed May 11 2005 - 13:15:26 EDT
This is additional followup to my original summary.
I got a followup email from Dr. Thomas Blinn who recommended AdvFS verify on the root domain. We also had a support call open with HP, recommending AdvFS fixfdmn.
After the usual struggles to boot-link a GENERIC kernel from the Tru64 5.1b CD and NHD7 CD (what a pain that is!) we ran /sbin/advfs/fixfdmn on the root_domain, which showed that the directory was indeed corrupt.
15:14:08 | INFO | Checking fileset root's directories.
15:14:08 | FIXED | Modified entry (1981.8019) to (1981.8040) in directory './cluster/members/member0/oldetc' (134.8001).
15:14:09 | FIXED | Added directory entry 'fixfdmn_lost_file_664.8031' to directory 134 in fileset 'root'.
Now I can delete the recovered rc.config and oldetc directory.
Thanks all for your help.
-----Original Message-----
From: tru64-unix-managers-owner@ornl.gov
[mailto:tru64-unix-managers-owner@ornl.gov]On Behalf Of Iain Barker
Sent: Friday, 06 May, 2005 16:54
To: tru64-unix-managers@ornl.gov
Subject: SUMMARY: Can't write to the rc.config file
Thanks to everyone who responded: Chris Eubank, Jeffrey Hummel, John Lanier, Franz Fischer, Eric Sisson, Michael Warchut
In summary, We're not running TruCluster on this system, but the use of /cluster CDSL's is standard for all Alphas on 5.1b with or without clustering. Other files in the the local /cluster/members/member0/etc directory could be edited, and all filesystems could be mounted, it was just the one rc.config file in this weird limbo state.
Nobody suggested what the root cause was.
The easiest solution was to boot from CD and mount the root filesystem manually, then rename the /cluster/members/member0/etc directory to oldetc and copy the files (except rc.config) into a new etc directory. Restore the backup copy of rc.config and reboot.
Of course, now I have a directory /cluster/members/member0/oldetc that can never be deleted because it contains a nonexistant file, but I can live with that...
-----Original Message-----
From: tru64-unix-managers-owner@ornl.gov
[mailto:tru64-unix-managers-owner@ornl.gov]On Behalf Of Iain Barker
Sent: Friday, 06 May, 2005 15:31
To: tru64-unix-managers@ornl.gov
Subject: Can't write to the rc.config file
Due to user error, the rc.config file on one of our Alpha 5.1b systems got deleted and the system was later rebooted so came up without any locla configuration.
I tried to recreate rc.config from a similar copy on another Alpha, but I cannot write to the file - it seems to be in some sort of limbo state.
I can rm the nonexistant file (it doesn't say 'No such file or directory" as it would if the file weren't present) but I can't recreate a file with that same name either manually or using rcmgr...
See console logs below - it's almost like the rc.config file is present in the filesystem (AdvFS with LSM mirroring) but not accessible. There aren't any processes running that would have a lock on the file though, as the system is currently booted single user.
INIT: SINGLE-USER MODE
fopen: /etc/rc.config: Invalid argument
# mount -u /
# ls -l /etc/rc.config
lrwxr-xr-x 1 root system 39 Apr 6 21:08 /etc/rc.config -> ../cluster/members/{memb}/etc/rc.config
# cd /cluster/members/member0/etc
# rm -f rc.config
# ls > rc.config
rc.config: cannot create
# rcmgr set HOSTNAME csgds1
# rcmgr get HOSTNAME
Error: Cannot open /etc/rc.config
#
Any ideas?
thanks.
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:50:18 EDT