Re: cache log write errors :again...!!

>
>
>
> On Wed, 2 Feb 100, Nikos Chrysos wrote:
>
> > I have configured my cache_dir to occupy 950 Mbytes (the partition is
> > a little bigger 1.1G)
> >
> > when the disk gets full (about 95-99%)!! occupying 1017Mbytes !!
> > i see write errors on the cache log going like this:
> >
> > 28) No space left on device
> > 2000/02/02 19:18:37| storeSwapOutFileOpened: Unable to open swapfile: /cache2/cache/02/21/000221F2
> > (28) No space left on device
> > 2000/02/02 19:18:37| storeSwapOutFileOpened: Unable to open swapfile: /cache2/cache/02/21/000221F3
> > 000/02/02 20:13:09| Select loop Error. Retry 10
> > FATAL: Select Loop failed!
> > Squid Cache (Version 2.2.STABLE4): Terminated abnormally.
> > CPU Usage: 1419.210 seconds
> > Maximum Resident Size: 0 KB
> > Page faults with physical i/o: 313783
> > Memory usage for squid via mallinfo():
> > total space in arena: 100928 KB
> > Ordinary blocks: 97037 KB 3614 blks
> > Small blocks: 0 KB 0 blks
> > Holding blocks: 12636 KB 55 blks
> > Free Small blocks: 0 KB
> > Free Ordinary blocks: 3890 KB
> > Total in use: 109673 KB 109%
> > Total free: 3890 KB 4%
> > 2000/02/02 20:13:09| storeDirWriteCleanLogs: Starting...
> > 2000/02/02 20:13:09| storeDirWriteCleanLogs: /cache2/cache/swap.state.clean: (28) No space left on device !!!!!!!!!!!!!!!!!!!!!!!!!!
> >
> > "CANNOT RESTART...............??!(my comment)"
> >
> > 2000/02/02 20:13:09| Finished. Wrote 0 entries.
> > 2000/02/02 20:13:09| Took 0 seconds ( 0.0 entries/sec).
> > 2000/02/02 20:13:12| Starting Squid Cache version 2.2.STABL
> >
> > But never comes up again.....
> > Is there something i am missing (there must be...:) )
> > I told it to use 950 and it uses everything it finds (leading to it's destruction..!!!!!!!)
> > I must say that the request rate is very big...(~80 req/sec...)
> >
> > Any help valuable
>
> Try a number lower than 950 Mbytes!!!!!!!!
>
> Start with 850, see if that works.
>
>
>
>

I 've tried 800 Mbytes ... but the same think happens ...the same errors
and then fatal , after fatal...
the strange think is that df shows me that 95% of cache1 is occypied when
the crash occurs...
Probably there could be some internal fragmentation...but could this be the reason..? and anyway should squid crach..?!
I could need any suggestions ..
thanks
nick

do dyou think that i should upgrade to new version...?
Received on Thu Feb 03 2000 - 05:29:13 MST