> Amos Jeffries schrieb: >> >> For specific config Markus, increase cache_mem. >> "cache_mem 0 bytes" will use 2-3 FD per request. When objects are pulled >> from memory that usage halves. > > What do you mean with specific config? > Funny thing is in my config cache_mem is set to 12288 MB. Just stumbled > now over this. Wrong syntax? I took the configuration from the Cache > Managers "Current Squid Configuration". But looking into "General > Runtime Information" it tells that it has above 12288 MB. Um. I meant to say 'your specific config', so others later don't try it randomly. But if you have it configured already and its showing up like that something is broken. Might be output display there. > >> But I agree its probably that FD leak you are hitting. > > Is there a workaround or ETA when this might be fixed(please take no > offense at this, it's just that the people paying me want to know things > like this)? Except checking the FDs and restarting Squid. I intend to look into it as soon as I have time, but don't really know the store code well and am also busy until after the Aus meeting. Looks like we will be discussing store there so hopefully I'll get some additional knowledge to throw at it after. Meanwhile anyone who can play around and shed more light on the issue would be very useful. Amos
