First thing to check: Is the AFS cache in its own, dedicated filesystem partition.
Second thing to check: When you say you are downloading the same file multiple times, is it exactly the same file, or is there anything happening to the file so that it changes (ie, some process rewriting the file, over and over, etc...even if the contents of the file stay the same) This can be tricky to solve, but I'm sure there's a solution. Good luck! Anne ________________________________ From: GALAMBOS Daniel <[email protected]> To: [email protected] Sent: Thursday, December 26, 2013 9:48 AM Subject: [OpenAFS] Apache and Openafs Hi, Our webservice built onto AFS, one node provide the static data, and works as a proxy for PHP. On this server, we have performance problem with AFS. This is a Debian Wheezy with afs client 1.6.1-3+deb7u1. So we got problems, i cloned the machine and started testing. I used apache benchmark (ab) against a simple html (w/o pictures or anything, just text). If i downloaded it 2000 times with 50 thread i got high load (18) and ~20 000 context switching in vmstat. The same file with 100 thread generated ~50 load. afsd is started with the following options: OPTIONS="-stat 10000 -dcache 10000 -daemons 6 -volumes 2048 -memcache -chunksize 18 -rxpck 750 -rxmaxmtu 1500 -prealloc 750" cache size is: 3065648 During the tests no other connection ( except apache status query 1/sec) was made, the same file was asked for 2000 times. After reading old messages on this list, i decreased the daemons to 3, which caused a very little improvement. Reading the apache status i noticed, that the apache spends 4-5 times more time in kernel space than in users pace. When the same file hosted on the local filesystem the avg response time decreased from 1800 ms to 50 ms and there are almost no load. Could you help me with this? Thanks, Dancsa _______________________________________________ OpenAFS-info mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-info
