Hello all...

Running Red Hat Enterprise Linux 3 on a 2.4 kernel.

The other day I went to query the RPM database (rpm -qa |grep SDD) and after about 30 minutes I noticed that I had no results and the command did not complete.

So I ran rpm -qa with no grep on a second terminal, and it hung, with no output.

Control C did not cancel the commands. viewing the processlist showed about 40 rpm queries running, using the following entries:

root 29389 28763 0 06:02 ? 00:00:00 /bin/sh /etc/cron.daily/rpm
root 29390 28763 0 06:02 ? 00:00:00 awk -v progname=/etc/cron.daily/rpm progname {????? print progname ":\n"????? progname="";???? }???? { print; }
root 29391 29389 0 06:02 ? 00:00:00 /usr/lib/rpm/rpmq -q --all --qf %{name}-%{version}-%{release}.%{arch}.rpm\n


None of them appeared to be going anywhere either (they stretched back for a while).

So i attempted a kill command on my rpm queries, with no results. I had to resort to kill-9.

I killed all the rpm commands that appeared in the process list too. They weren't doing anything, I thought maybe the original one hung up and they all got stuck. But still, rpm queries keep failing.

What's going on with my rpm database?

on an aside, I noticed too that my locate database is more than 8 days out of date. Not sure when that stopped working. Could it be related?

Thanks in advance.

--zigg

You could try to rebuild the rpm data base using the rpm command:

rpm --rebuilddb

See:

man rpm

--------------------
Steve Stites

P.S. Another possibility is that there may be a lock file for the rpm data base which did not get deleted properly. I don't remember where the rpm data base lock file is. It might be in /var or perhaps /etc. You could try to find it with these commands:

find /var -iname "*lock*"

find /etc -iname "*lock*"

I found a lock file here: /usr/lib/rpm/rpmdb_deadlock
but that file was created in April of 2006.
I didn't see any lock files pertaining to rpm in /var or /etc. Most of the lock files dealt with databases, the rest were rsync, nfs, etc.

I am running the rpm --rebuilddb -v now, but it's been 10 minutes and there has been no output from this command. I don't think it's actually doing anything, but I'll give it some more time and see.

This is our main production server, so I may reboot it this weekend to see if that clears up whatever the issue is.

Thanks for the help.

--zig

Something is wrong. Rebuild shouldn't take anywhere near that long.

I took a look at my CentOS system which is rpm based. It has a file called /usr/lib/rpm/rpmdb_deadlock even though I am not currently booted into CentOS.


--------------------------
Steve Stites

Yes, this much I gathered.

Any troubleshooting suggestions?

When you take down the system I suggest that you run fsck against the unmounted partition containing the rpm database.

-------------------
Steve Stites

Apparently, this was the problem:

Found it here: http://www.linuxforums.org/forum/red...just-hang.html

Thanks for the help though!

remove cached DB
rm -rf /var/lib/rpm/__db.00*
rpm --rebuilddb

for advanced debugging (while hanged )
strace rpm -qa