On 7/06/2015 4:49 a.m., Mike wrote: > Running Scientific Linux 6.6 (based on CentOS), compiled squid 3.5.5 > with no errors. > Issue happens with both squid 3.5.5, and the newer 3.5.5-20150528-r13841. > Starting squid service and running for a while, any attempt to access a > website like www.nwfdailynews.com causes squid to crash and restart. The > associated squid log entries from the same time squid crashes, seems to > be several "TCP_SWAPFAIL_MISS" at that time: > > 1433608644.053 117 192.168.2.110 TCP_SWAPFAIL_MISS/200 17246 GET > http://launch.newsinc.com/77/css/NdnEmbed.css - > HIER_DIRECT/184.51.234.134 text/css > SWAPFAIL is either a disk I/O error, or the file has been erased from disk without going through Squid making the cache index entry pointing nowhere. >From a service point of view thay are nearly harmless, though possibly a sign that you need to check the HDD integrity. > Other normal use including facebook and other website do not have any > problems, so it has to be something relating to this site setup causing > the crash. We are using this in a testing server for now but hoping to > roll out to production level used by a few thousand people. > That domain has bigger problems than what you see. redbot.org shows its origin server is not generating Vary headers properly. Which means visitors to the website that go through a proxy (any HTTP proxy) will be getting randomly broken responses. Furthermore, it appears to be served up to the public through a CDN built from Squid proxies. So that is affecting *every* visitor. Amos _______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
