Red Hat Bugzilla – Bug 101411
--rebuilddb and -qa fails to create __db* files
Last modified: 2007-04-18 12:56:28 EDT
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686) Gecko/20030530 Galeon/1.3.5
Description of problem:
As mentioned in the email@example.com list, I cannot create indexes for rpm.
I've changed LANG=en_US and also tried with LANG=C (and unset LANG), with no effect.
I cannot figure out how to pipe the results to a file for attachment:
rpm --rebuilddb -vv > hideously_slow.asc
still spews to the console. Can you remind me how this is supposed to be done?
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Get a ThinkPad T30, install RH8, and up2date and ximian evolution updates.
2. rpm (anything, rebuild, install, upgrade etc.)
3. practice singing "Anticipation..."
Actual Results: rpm --rebuilddb takes 8 minutes on a freshly-installed 1.2GHz
laptop, but takes half that time on my ancient dual-PIII 450. ?
And when I am done, there are no __db* files in /var/lib/rpm...
Expected Results: 'rpm -qa' or 'rpm --rebuilddb' do not provide the __db* files
As soon as someone can remind me how to pipe the results to a file, I'll attach
to this bugreport.
Created attachment 93301 [details]
results of rpm --rebuilddb -vv
1) I have ~1000 pkgs, /usr/bin/time rpm --rebuilddb -vv took 1:45.31 minutes
on 600 MHz dual i686, SCSI.
2) --rebuilddb does not recreate __db* files, any other command
run as root should create, but existence of __db* files is not
really a problem.
3) what version of rpm are you using?
Hmmm, so --rebuilddb on laptop is slow.
Can you provide "cat /proc/cpuinfo" from laptop?
Can you also check /usr/bin/time rpm -qa on your two machines?
I did the sanity checks, and here's what I get:
1) I have 761 packages, and I get the following:
28.90user 4.08system 2:51.33elapsed 19%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (500major+13298minor)pagefaults 0swaps
The most annoying part about this is that while it is running, all other
applications (evolution, terminals, konqueror, etc.) grind to a halt (almost
making me think the machine is hung or out of disk). But you can see that
there's plenty of cpu to go around (Mobile Intel(R) Pentium(R) 4 - M CPU 1.80GHz).
2) There are no __db* files, even after root doing 'rpm -qa', 'rpm --rebuilddb',
'rpmbuild --rebuild (SRPM)', etc.
Results of 'rpm -qa' timing test: laptop was faster than old machine, but -qa is
not what is taking so long. It is -Uvh, --rebuilddb, -e and -i that are taking
OK, then you're probably slow (but not exceptionally so) because
of disk writes, that's what differentiates -U/-e/-i/--rebuilddb from -qa.
The other processes stopping is (my guess) probably the same cause,
there's nothing magic that rpm is doing, I suspect that any output
intensive program would have similar effect on other processes, possibly
because of swapping.
Try running vmstat 1 in a window to eyeball your disk I/O. Running top
might give you a handle on swapped too.
SO what happens if you do, say,
tar czvf /tmp/xxx.tar.gz /usr
while monitoring. Do disk writes cause slow response?
Hmm, this machine is swapping like mad. Could it be the new-and-improved kernel
I'm gonna try booting into the old kernel and see...
Regardless, looks like you're of the hook regarding RPM ;^)
Thanks for the help.