Help, 'bad swap_file_number'

Hello,

I am fairly new to this list and apologize deeply if my question has
already been treated, I did look over the FAQS and the online search. If I
am redundant please just say that the above sources have my answer and I
will gladly go digging again.

I have been using Squid satisfactorily since squid-1.NOVM.16 (I think). As
our ISP has grown my squid has been dying with more frequency. I thought at
first that this was a FD problem but the logs didn't seem to indicate this.
The 'squid deaths' have become so frequent now (every couple of days) that
I decided to seek the counsel of this list.

The key entry in my 1.NOVM.20 cache.log is:

...
1998/04/25 00:21:34| Took 4 seconds (3445.5 objects/sec).
1998/04/25 00:21:34| store_swap_size = 160651k
1998/04/25 00:21:40| file_open: error opening file
/usr/local/squid/cache/08/5D/000035D8: (2) No such file or directory
1998/04/25 00:21:40| storeOpenSwapFileWrite: Unable to open swapfile:
/usr/local/squid/cache/08/5D/000035D8
1998/04/25 00:21:40| WARNING: storeOpenSwapFileRead: bad swap_file_number
1998/04/25 00:21:40| Swapfile open failed
FATAL: file_write: bad FD
Squid Cache (Version 1.NOVM.20): Terminated abnormally.

... then it respawns over and over again.

I am using Linux 2.0.32. The cache is a softlink to another partition
(exclusive for squid use). Furthermore, there is never another cause of
squid termination other than the above error.

Any ideas? A bad sector on partition? No idea really, just thinking.

If you do got an idea as to what I am up against I would really appreciate
a pointer. Also, if this is a common ailment and already well discussed, I
apologize. I have just gotten so use to restarting squid with a freshly
formatted partition (my only solution up till now) that I even have a
script for it.

Many thanks,
Alex Panagides
Ceara, Brazil
Received on Fri Apr 24 1998 - 21:25:35 MDT