I have an Squid proxy configured as a reverse proxy that serves about 10 web sites via SSL. One of the web sites randomly becomes unavailable and the only way to make it work again is to restart the squid process. The following error message in the cache.log is repeated lots of times when the problem occurs: 2007/09/28 10:06:09| fwdNegotiateSSL: Error negotiating SSL connection on FD 148: error:00000000:lib(0):func(0):reason(0) (5/-1/104) 2007/09/28 10:06:09| TCP connection to 1.2.3.4/443 failed I check the backend web server directly and it is alive so technically the cache_peer is available, but squid doesn't seem to think so. The other odd item is in the access.log, normally I would expect to see a "TCP_MISS/200" and the matching cache_peer name at the end of the request line : 1190984166.734 32 11.12.33.44 TCP_MISS/200 280 GET https://www.domain.com/c3.gif - FIRST_UP_PARENT/website_ssl - But when the problem happens, I get "TCP_MISS/503" and NO matching cache_peer: 1190983242.734 32 11.12.33.44 TCP_MISS/503 280 GET https://www.domain.com/c3.gif - NONE When this event is happening, all the other web sites are just fine. It seems to be specific to this web site. I restart the squid process, and the site resumes working. Any ideas? Thanks.
