On 23/10/10 21:08, Dmitry Sinyavin wrote:
Hi All! I'm running 3 squid hosts (not siblings, just dns round robin), all 3 have the same config file (i mean absolutely no difference). But while two of them are running smoothly, 3rd have a problem - although it is serving requests normally, its store.log file is growing by approx. 1Mb/sec. File mostly consists of lines like: 1287820122.900 RELEASE -1 FFFFFFFF AFA36A32B9A4BE25DF197B156F9EDD4C ? ? ? ? ?/? ?/? ? ? 1287820122.900 RELEASE -1 FFFFFFFF B7978B41EE7BF36B5049B7D1B088B2C2 ? ? ? ? ?/? ?/? ? ? 1287820122.900 RELEASE -1 FFFFFFFF C298A1D274A53F38B0FC630CE0380599 ? ? ? ? ?/? ?/? ? ? And so on, 80Gb of them (well, sometimes there's normal lines too). Any thoughts? I made it "config_store_log none" in config for now, but the situation is scaring me a little, those servers are running a large project ;)
store.log is a journal of events happening in the storage. I suspect this are the in-memory cache events being logged.
In most uses store.log not needed at all. You could add "cache_store_log none" to your config.
You may also want to re-asses how often "squid -k rotate" gets run on a busy system. Part of the operation alongside log rotations is swap.state journal compaction. This one is likely to be more of a problem if it gets up into the multi-GB.
Amos -- Please be using Current Stable Squid 2.7.STABLE9 or 3.1.8 Beta testers wanted for 3.2.0.2
