Matus UHLAR - fantomas wrote:
On 07.11 20:18, forgetful tan wrote:
My squids ran into trouble after it startup for some period of time.
Squid stop accepting new connections. When I login to that machine and
ran df, it show out that there's 100% space usage in the cache_dir .
Read: http://www.squid-cache.org/Doc/FAQ/FAQ.html#toc4.14
maybe you configured cache_dir too big.
I have 27g of disk space, I config that cache_dir to 8g .
I tried to stop squid and re-newfs . But I just can't umount it !
There's probably process having opened something on that filesystem
(e.g. it's current directory is on it)
The only way I can fix it is to reboot the machine to let it be fscked.
And after the reboot, I df again and got about 20% disk usage in the
cache_dir .
did you "rm -rf" everything in that directory?
surely no!
I met the same problem in FreeBSD from 4.3 to 4.9 with
cache_dir of UFS+SoftUpdate?? I check the google for UFS problem and got
some advice on changing the time optimize to space optimize. But I just
don't think that's the problem I met, coz if all the blocks are used up,
then it can't be fix after a simple reboot (with a fsck ?), right ?
there may be process having open files which were removed, but as long as
they are open, they still take space on the disk. After reboot they get
physically removed.
Where do you store your logs?
Another problem is that when I tried to umount that directory, the
umount process just hung up, even can respone to the ctrl + c signal !
It seems like a FS's problem.
