I have encountered a likely bug which I cannot find any exact duplicates
of, but I thought I could mention it to you before submitting a bug
report, to see what you think.
I've had problems under 2.6.32-5-amd64 on both a HP Proliant DL120 G7 and
a Supermicro X9SCL-F. The Supermicro box runs a Intel(R) Xeon(R) CPU
E3-1270 V2 @ 3.50GHz.
The problem is that processes on the machine stall for a few seconds to a
few minutes for no obvious reason. The stalls can include lightweight
processes such as vim or alpine, sometimes when writing a file, sometimes
just editing text (I've had four stalls already this far into the email).
I run sync from cron periodically and dmesg shows the occasional complaint
about it hanging:
INFO: task sync:32577 blocked for more than 120 seconds
.. trace ..
Other sessions continue to respond while this is happening. I read in
other bug reports to run echo w > /proc/sysrq-trigger which I have done,
the output is here:
The first sysrq trace is just testing, the second was while a process
(this alpine instance) was stalled.
I retired the HP machine because I assumed this problem was caused by
hardware problems. I didn't notice the problem on the Supermicro as a
fresh install, but now that I have all my software on here it is also
playing up. The machine is not particularly loaded (load averages about
1.5). Some of my software downloads a few dozen small gif files per
minute and runs imagemagick on them, so there is a little IO but it is
pretty trivial really. There is plenty of contiguous space left on the
Filesystem Size Used Avail Use% Mounted on
/dev/sda2 1.8T 98G 1.6T 6% /
Any thoughts on where I should go from here?
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact firstname.lastname@example.org
Archive: alpine.DEB.email@example.com t.com.au">http://lists.debian.org/alpine.DEB.firstname.lastname@example.org t.com.au