Hey Roie, From the output I assume it’s a dns resolution issue. In the past I remember that Docker was updating the hosts file with the relevant names but it’s not working the same way now. Currently Docker is using a local network dns service which is being accessed via 127.0.0.53. From I remember Squid is resolving the icap service name only at startup or reload. Lately Alex published a testable patch that might fix specific issues with icap services which are resolved by dns. ( sorry I don’t remember the bug report) I assume you can try to test this patch first. If these services are static to some degree you might be able to create a script that updates the hosts file and reload squid on each change. When using the hosts file it’s possible that some issues will disappear.
You might be able to use tcpdump from either the host or the container side to capture traffic when these goes down. Depends on your preference of debug level you might even be able to debug specific debug_options like for ICAP services If you really need help with a diagnosis and a solution you might be able to use Alex and the measurement factory. All The Bests, Eliezer From: squid-users <squid-users-bounces@xxxxxxxxxxxxxxxxxxxxx> On Behalf Of roie rachamim Hi, Our setup includes squid that runs in docker container with several ICAP servers in additional containers. From time to time we see in cache.log the following messages: We're trying to understand why it happens to all ICAPs at once. This happens in 4.14 and in 5.0.4 Any thoughts about what might cause this ? Many Thanks, Roie |
_______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
