[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: ext3 device reported to be 100% full, but we do not know where? - solved



Hi Stephen,

> > ok, then at least lsof should report such files.  Actually I forgot
> > something: Even after all processes holding this file handle, the
> > storage was never freed, only after umount/mount!
> 
> Can you reproduce this?  In every case where this has been reported in
> the past, the user has eventually discovered that there was indeed
> some process somewhere holding a file open.  If a "df; umount; mount;
> df" shows a reproducible disk space leak then there's an orphan leak
> somewhere.

It happend at least 2 times with the something like the following script
'script':

#!/bin/sh
someprogram &

$ nohup script

Where a cronjob deleted the nohup.out every 15 or 30 minutes. 

"someprogram" spawned child processes in a loop, where the child processes
printed to stdout (hiere nohup.out). The whole thing was just a test
scenario, just for the case you call this a strange setup, it is.

Even after successfully killing someprogram and all its child processes,
the storage was not freed. I can do more tests probably in one or two
weeks. Other than up to yesterday,  the standby server is now in
productive use, where I should not flood any of the filesystems.

But now, where it comes to such details: We are using Debian Woody with
this kernel:

$ cat /proc/version
Linux version 2.4.19-pre8 (root pima) (gcc version 2.95.4 20011002 (Debian
prerelease)) #4 SMP Sun May 5 15:00:54 CEST 2002

The machine is 2 dual PIII, if that is important for some reason.

	Michael

-- 
Hostsharing eG / Boytinstr. 10 / D-22143 Hamburg
phone+fax:+49/700/HOSTSHAR(ing) (= +49/700/46787427)
http://www.hostsharing.net: host, where YOU make the difference





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]