In one part, i was wrong, Today the 32-bit squid 2.6 is crashing often, too. I'll find out why (I have modified like bug #1655 my squid, so I now can generate core dumps under an actual linux) and will send you the trace, when I've got one. Martin -----Ursprüngliche Nachricht----- Von: Stolle, Martin [mailto:Martin.Stolle@xxxxxxxxx] Gesendet: Dienstag, 4. Juli 2006 10:06 An: squid-users@xxxxxxxxxxxxxxx Betreff: Questions or problems in transition from SQUID-2.5STABLE14 to SQUID-2.6STABLE1 I tried to upgrade two of our main proxies from squid-2.5.STABLE13/squid-2.5STABLE14 to squid-2.6STABLE1 One is a SuSE Linux 10.0 - x86-32, the other is a SuSE 9.3 - x86-64. Both are connected to a hardware loadbalancer and have about 3000 clients and about 100-150 requests per second each. While the 32 bit machine is running stable, the other one (64-bit) restarts often with Segmentation Violations. Both are configured with nearly the same parameters (with the exception of optimization and 32/64-bit) Before that, both machines ran stable for months with the epoll patches from devel.squid-cache.org. I switched back on the 64-bit machine, and the squid-2.5.STABLE14 runs stable again. Is it possible, that there is still a 64-bit Problem in Squid-2.6STABLE1? I couldn't get a core dump of the Squid 2.6.STABLE1 - x86_64, perhaps because it runs multithreaded (with aufs). --- The other effect: The cache digests aren't longer working, both with the 32 bit and the 64 bit system, also I enabled them with -enable-cache-digests. In the cache log, I can find an error like "temporary disabling digest from vproxy.. (not found)", in the access.log the digest request is responded with a 404 error. Is there any change in the syntax of cache digests in squid.conf, so that this could be expected? Thanks in advance, Martin
