[Gllug] kill -9 does not work

[Vincent AE Scott]

On Tue, 26 Jun 2001, Xander D Harkness wrote:

> we are using level3 for some of our hosting (well racking and 30 feet of
> cable to our provider) and they have killed our connectivity  - dunno
> hacksawed through a cable or something - but anyway we have no external
> access to my lovely production boxes - so I took the opportunity to errrrrm
> yes err reboot....

i saw this happen with early RPM-4 bins, FWIW.
IIRC it *looked* like a synchronisation problem, either fork/wait or
thread cond-wait locking, or even file locking, cant remember which
exactly.  extremly annoying, but since upgrading to a more recent RPM
binary, it's all been working fine.  that was on a MP machine, so *maybe*
it was a race condition that was more evident than normally on a uni proc
boxen.


> I should mention also that the command ls would hang as well.

not uncommon.  in my experience usually due to either one mount point
blocking while the kernel attempts some stat style stuff on the root inode
for the device.  or, when user names cant be mapped to UID's, NFS hangs,
NIS timeouts etc.

> 
> All is new and well :-) my server is up and happy again.
> 
> I would have like to know what the output and problem was.

# strace -p `pidof rpm`
# lsof -p `pidof rpm`

*if* rpm fork's a child, then check what they are all doing.  i know that
im telling you how to suck eggs with that, but usefull for those that dont
know about it.


> Cheers
> Xander
> 
> 

-- 


PGP key:  http://codex.net/pgp/pgp.asc


-- 
Gllug mailing list  -  Gllug at linux.co.uk
http://list.ftech.net/mailman/listinfo/gllug