[PLUG] Time to downgrade?

म.हा.सा.ग.र o.s.h.o at guruvision.com
Wed Jan 7 19:59:26 IST 2009

On Wed, Jan 7, 2009 at 7:48 PM, Shridhar Daithankar
<ghodechhap at ghodechhap.net> wrote:
>> =-=-=-=====-=-=-=-=-=-=-=-=
> Hoping that the hang lasts for several seconds, rerun the strace, find the
> offending fd i.e. 10 in the above strace(first argument to the read, which is
> repeatedly unavailable) and see what file it is trying to open.

while this is happening it is happening only for a single large kmail
folder... even if i am surfing somewhere else...

will regenerating index of that folder help?

well I am guessing it might... just hoping to hear an echo ala Konika
... Sakura :-)

> You can dereference the fd -> file name in /proc/<pid>/fd directory.

Yes in my case also it is a socket...

>> I am also wondering if this has got anything to do with *TSC clock
>> source unstable*  message that i am getting for last few days...
>> which maybe related to cmos battery?  But can that battery cause a
>> problem when i am on M.S.E.D.C.L supply ?
> Once the machine is booted, cmos battery doesn't come into play AFAIK. Its
> main job is to keep the clock running while the machine is off. When the
> machine is running linux, the OS keeps updating the time.


this is an interesting post on TSC matter... and i am wondering about it now...

> You could try syncing with a time server(which is as simple as running ntp
> client. No config required. Highly recommended.) Changing cmos battery is
> advised as well.

no ntp... probably looks like this has nothing to do with cmos battery..

>> anything else to look for?
> Find the offending file. kmail is spending lot of time trying to read from it.
> It should give some clue. In my case, the fd 10 is a socket.(this is weird. I
> could not get any info. on what the other end of the socket is, even with lsof
> and netstat.)

same for me... not that i understand much of netstat etc...

Another thing to check could be a rootkit (thinking of insanely in-secure :-)

More information about the Plug-mail mailing list