On 9/2/19 6:53 PM, service@xxxxxxxxxxxxx wrote: > I’m wondering how we can implement rules inside Squid 4.8 that if a > parent cache_peer is dead at the time of the proxy request, squid > automatically bounces it to another parent cache_peer instead of the > connection just timing out and not fulfilling the request due to the > current cache_peer being offline or otherwise unavailable. Basically > what I want to do here is to have squid automatically choose any working > cache_peer parent instead of the request just timing out or dying. If there is enough time (see connect_timeout and forward_timeout), there are other destinations to try, and a few other special conditions are met, then Squid should automatically re-forward the request to the next peer or origin server upon hitting a connection timeout with the previous peer or origin server. For details, see https://wiki.squid-cache.org/SquidFaq/InnerWorkings#When_does_Squid_re-forward_a_client_request.3F If your Squid does not re-forward, check the above-mentioned options and the wiki page to determine why. If all those preconditions are met, but your Squid still does not re-forward, file a bug report. Posting the corresponding access.log records and compressed cache.log (with debug_options set to "ALL,2 44,5 17,5" or higher) collected while reproducing the problem using a single HTTP transaction may help in triage. Alex. _______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
