Squid behaviour when a cache disk fails

From: Charles Bruneteau <charles.bruneteau.sysicom@dont-contact.us>
Date: Thu, 02 Sep 1999 18:55:06 +0200


As the LRU gets very slow, we are planning to add some hard disks to our squid box (2.2stable4, solaris 2.6). Actually we have 2 disks of 9 Gb each which are mirrored.

We wonder what would be the behaviour of squid if a disk (without mirroring), or a controller, starts to fail.

As a test we tried to remove one of the configured cache directories. Some messages such as

        1999/09/02 17:45:12| storeDirClean: /cache3/00/02: (2) No such file or directory

appeared in cache.log, but the service was still provided to the clients, that is the clients were receiving the objects they requested.

However won't squid hang in such conditions, under heavy load or after a certain amount of time ? If not, we would have time to reconfigure squid / change disk ...


Charles Bruneteau
Received on Thu Sep 02 1999 - 11:08:42 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:48:14 MST