Hello folks, I would like to get some opinions about the following mishap with Squid. Out of sheer curiousity I installed Squid 2.2DEVEL3 on Friday morning to found how/whether it works. As it seems, the ICP communication between my cache (2.2) and the peers (2.1 & 2.0) got somehow broken, because the peers started to fetch misses massively, resulting in a flood of error messages about my cache not forwarding requests. The quick first-glance analysis shows that 2.2 returns ICP_HIT for just about any URL it has ever heard about; after a TCP_REFRESH request from the peer it finds out that the URL is stale, yielding TCP_REFRESH_MISS and an appropriate error message. Is that an intended behaviour or a misfeature that will disappear as 2.2STABLE makes its way to the caching public? I understand that this would not be a point in a homogenous Squid 2.2 environment (due to the ability to retry failed/disallowed requests), in a heterogenous mesh however this would be an issue. Thanks for any comment about that. JAR
