Thanks a lot Todd
As Per IT the server used to go to Hang state and they
Had to reboot to get it back alive.
The problem started happeninig after
we moved cvs repository filesystem from
SAN(attached storage) to NAS(filer)
IT folks say that we are hitting a known NFS bug
With this old version of Red HAT.
Yesterday I had to clear our all locks left over by the
Crash and then I got errors for tag/rtag.
I panicked seeing that error message, but
LUCKILY when I tried the tag/rtag second time
It worked!
Best regards
pranab
-----Original Message-----
From: Todd Denniston [mailto:address@hidden
Sent: Friday, October 05, 2007 7:19 AM
To: Pranab Panda (pkpanda)
Cc: info-cvs list
Subject: Re: Serious cvs tag/rtag issues Please help
Pranab Panda (pkpanda) wrote, On 10/05/2007 01:17 AM:
Hi All
In last couple of days our cvs server has gone done few times,
You mean like power outage/hardware fail???
UGH.
Checkin and checkout work fine, but
since this afternoon I see errors with cvs tag and rtag commands.
They both fail with following error
cvs [rtag aborted]: could not open lock file
`/auto/cvsroot/repository/Files/,file12,': File exists
I am clueless how to fix this.
Please help if know whats the issue
this is very critical!
NOTE: cvs --version
Concurrent Versions System (CVS) 1.11.2 (client/server) On Linux
2.4.21-27.0.1.ELsmp #1 SMP Mon Dec 20 18:47:45 EST 2004 i686
i686 i386 GNU/Linux
Thanks in advance
pranab
Suggestions:
1) make a backup[1] of what you have... immediately.
2) if it was power/hardware fail, A) you should get new UPS/hardware, B)
probably fsck the drive with the repo on it.
3) `locate check_cvs; locate validate_repo` one of these should be on
the system in a cvs/contrib directory (assuming a nice Linux distro),
Otherwise you can beat the .in file[2] into shape.
4) `perldoc check_cvs || perldoc validate_repo`
5) use one of check_cvs || validate_repo on your repository to see what
other problems you have.
6) with the ",filename," files, I _think_ you should see if there are
usernames on them (if you are using pserver there might not be real
ones) and if so ask those users to finish their broken commits.
7) if you think it appropriate, make a new backup.
8) ask the group again about help with the new problems if needed.
9) upgrade to a much newer CVS/verify RH has updated for all the
security problems found since 1.11.2.
Good luck.
[1] http://ximbiot.com/cvs/manual/cvs-1.11.22/cvs_2.html#SEC24
[2]
http://cvs.savannah.nongnu.org/viewvc/*checkout*/ccvs/contrib/validate_r
epo.pl?root=cvs&content-type=text%2Fplain