Hi Amos , thanks for reply. But I think using large rock will let the memory cache size be >32 ??? am I correct ? I mean although im not using rock store , but I was thinking will have memory size >32Kb So , I keep the aufs size which I assume will help me and will let me break the 32kB limitation .. but it seems not like I was planning !!!! So m my previous config was using aufs with large rock needed in memory size which is not what I was planning ?!!!! can you plz explain me why my previous config don’t have luck with memory >> 32KB?? So for now , as u told me I removed all aufs drivers and added only rock dir as below : cache_dir rock /ssd1 90000 max-size=32768 swap-timeout=350 max-swap-rate=350 cache_dir rock /ssd2 90000 max-size=32768 swap-timeout=350 max-swap-rate=350 cache_dir rock /ssd3 90000 max-size=32768 swap-timeout=350 max-swap-rate=350 cache_dir rock /ssd4 30000 max-size=32768 swap-timeout=350 max-swap-rate=350 And only mean size I can see now is 18 Kb ??!!!! Why ??? I had tried another trial , I put memory_cache_shared off , squid cant be started all all kids is being killed and stared again and so on ! Sample of logs : FATAL: Received Segment Violation...dying. 2014/11/16 09:18:28 kid1| Closing HTTP port xxxx 2014/11/16 09:18:28 kid1| Closing HTTP port xxxxxx 2014/11/16 09:18:28 kid1| storeDirWriteCleanLogs: Starting... 2014/11/16 09:18:28 kid1| Finished. Wrote 0 entries. 2014/11/16 09:18:28 kid1| storeDirWriteCleanLogs: Starting... 2014/11/16 09:18:28 kid1| Finished. Wrote 0 entries. 2014/11/16 09:18:28 kid1| Took 0.00 seconds ( 0.00 entries/sec). CPU Usage: 8.235 seconds = 8.049 user + 0.186 sys Maximum Resident Size: 381440 KB Page faults with physical i/o: 1 2014/11/16 09:18:30 kid5| Current Directory is /root 2014/11/16 09:18:30 kid5| Starting Squid Cache version 3.HEAD-20141105-r13687 for x86_64-unknown-linux-gnu... 2014/11/16 09:18:30 kid5| Service Name: squid 2014/11/16 09:18:30 kid5| Process ID 2925 2014/11/16 09:18:30 kid5| Process Roles: worker 2014/11/16 09:18:30 kid5| Process Roles: worker 2014/11/16 09:18:30 kid5| With 131072 file descriptors available 2014/11/16 09:18:30 kid5| With 131072 file descriptors available 2014/11/16 09:18:30 kid5| Initializing IP Cache... 2014/11/16 09:18:30 kid5| Initializing IP Cache... 2014/11/16 09:18:30 kid5| DNS Socket created at [::], FD 10 2014/11/16 09:18:30 kid5| DNS Socket created at 0.0.0.0, FD 11 I wish you help me with a good squid.conf file that can help me in saving bw as I can ! regards -----Original Message----- From: Amos Jeffries [mailto:squid3@xxxxxxxxxxxxx] Sent: Saturday, November 15, 2014 10:07 PM To: Ahmed Allzaeem Cc: squid-users@xxxxxxxxxxxxxxxxxxxxx Subject: Re: squid cache Large rock with aufs optimization for bandwidth saving -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 17/11/2014 2:24 a.m., Ahmed Allzaeem wrote: > Hi Amos , thank you > > Really sorry to ask you , wts needed to be done to use largerock > support ? Example 5GB rock cache; cache_dir rock /ssd1 5000 > > I mean wt wring im doing ? > > Also about the cace_mem directive , as I understood this value per > worker or per process. And here im not using rock , so its not shared > , I have about 32 Gram total. http://www.squid-cache.org/Doc/config/memory_cache_shared/ You are using workers, so it is shared unless you configure: memory_cache_shared off Note this: "Currently, entities exceeding 32KB in size cannot be shared.". I am not sure if that was fixed when large-rock support was added. Should have been, but maybe not. For most of the per-worker settings you have are undone by the global settings at the end of the config between http_port and WCCP config. > > Can you correct nme about setting cache_mem directive ? also guide me > wts needed to enable caching large object size like 512k - 1 M or even > 2 M > Try "memory_cache_shared off" at the top of your config. Amos -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (MingW32) iQEcBAEBAgAGBQJUaD7vAAoJELJo5wb/XPRjDfYH/1SsTIG4RUrU5NXB7C4Vojva iZdt9hPLW7QTkUWjjB3tr6Y84I4DRCd9/N6h1LBlDY3lNmjhsYQEwCEUmm6Dm2Uo hj1P1Nvn6cBMOjDZswFrz3imo37UAnvIUAqqwMC2SJF3uglvgtrd8IdeFVPRIoSR bgjsiZB1yAx+SBCEOPs2sEhpqdi0215FeiF/ujTS5A7JeEPS+jPVcZoYT/xRoe9L oDXkITcfBlOa/0bY1lo5LAqPHmVDh5BZxVn8ToMMvki21J9UscRdvT04JVoc6iqP Uouo9VxZJDJfxF0BtK4wJSG1nKtcTiFYrBqK862L2c7zwYPB9WIPpUflKRx/aFQ= =82cL -----END PGP SIGNATURE----- _______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
