The store url mismatch stuff means an object wasn't fetched from cache but the client won't notice the difference - it'll just be a miss. I've seen this creep up when storeurl rewrite rules change and generate different backend names for objects. As for COSS, I don't suggest running it on Windows. I never tested it. As for low-priority transfers - if your OS supports setting the TOS on an already-established TCP connection then it wouldn't be difficult to patch Squid to reset the TOS for that socket mid-flight on quick_abort. 2c, Adrian 2008/8/20 chudy <chudy_Fernandez@xxxxxxxxx>: > > even using the 2.7 stable 4 version(binary for windows) with newly created > swap files still the same. i've been using storeurl and aufs feature since > from the squid head. now that im trying to use coss this warnings came up. > > Henrik Nordstrom-5 wrote: >> >> sön 2008-08-17 klockan 20:41 -0700 skrev chudy: >> >>> one thing i've seeing Warnings about failed to unpack meta data that i've >>> never seen in aufs. >> >> Did you wipe your cache when changing the file size api? >> >> 32-bit and 64-bit caches may be incompatible.. >> >> Regards >> Henrik >> >> >> > > ..or maybe storeurl is not final. bec. storeurl mismatch when the content > is store in memory and revalidated. but on the second thought no need to use > storeurl on smaller objects since speed is our concern. bec. this objects > usually give warnings about meta data are smaller objects. i've tried > storeurl_access deny smaller_content that are smaller than > maximum_object_size_in_memory it seems works fine. > > but still i need confirmations. > > on the other thought i've been thinking if the objects being canceled by the > clients, i want to continue downloading in squid but in lowest priority of > bandwidth... is it possible? or any workaround to make it happen? > quick_abort_max to -1 (correct me if i'm wrong) uses same bandwidth. it will > be total congestion if these files are videos. its really nice if it will be > on lowest priority and what makes ever better if the client retry to > download the priority back to normal. > -- > View this message in context: http://www.nabble.com/Mingw%28patch-for-long-file-pointers%29---with-large-files-tp19025674p19070570.html > Sent from the Squid - Users mailing list archive at Nabble.com. > >
