On 13/09/2016 8:26 a.m., --Ahmad-- wrote: > amos thank you so much for reply > > > how can i correct the patterns below ? > Remove the ([^.]+.)? and ([^.]+.|) parts. Replace the . where you want to match a '.' with \. > > im not sure if the other patterns are doing the game > > but so far i don’t have disk increasing at all > > only the disk start increase when only i add : >>> refresh_pattern ([^.]+.)?(download|(windows)?update).(microsoft.)?com/.*.(cab|exe|msi|msp|psf) 4320 100% 43200 reload-into-ims >>> refresh_pattern ([^.]+.|)(download|adcdownload).(apple.|)com/.*.(pkg|dmg) 4320 100% 43200 reload-into-ins > > but i still don’t have HITs > You have many CLIENT_REFRESH/20x. Which means cached content was found, but the client required it to be refreshed. When revalidated the server returned an entire new object. With the above refresh_rattern, the client may have required the cached data to be reloaded (Cache-Control:max-age=0) and the reload-into-ims converted it to a CLIENT_REFRESH from a MISS. > im not sure where should i tune ? There is not much you can do when the server returns whole objects in response to revalidation requests. The client is requiring that only the latest copy be used, and the server is insisting that the object has changed. Since this is an OS update process it is exremely unsafe to make assumptions yourself about anything being differently cacheable than what the server is insisting. It is entirely possible that each response is encrypted and hashed independently - so actually is different. > > i also added stip query terms to off > > > not sure what should i do next ……. all i need is windows updates go ok , I’m not interested with other caching > With query-terms in the access.log you can better see what requests should be MISS and which ones have a chance of being HITs (or near-HITs). Once you can tell which requests actually should be HIT-able you can then look at the messages going through (in cache.log with debug_options 11,2) for those requests to see what else might be done. Amos _______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
