Hi ,
I am using Squid 2.1Patch3 on Solaris 7.0/X86 with
PII450 512MB - 2 CPU
RAM-256 MB , 2 7500 RPM SCSI II HDD- 4 GB each.
I am using Polygraph 1.0p7 with(patch) but squid not
able to handle more than 75/req sec. ( --dhr 55% and
--rep_cachable 80p same as bakeof)
I have tuned TCP( Solaris 7.0) for following
parameter...
tcp_time_wait_interval 60000
tcp_conn_req_max_q 16384
tcp_conn_req_max_q 16384
tcp_recv_hiwat_minmss 24
FD is set to 12288.
Even by increasing RAM up to 1 GB , I don't find any
improvement in performance.
I get the following error in my cache.log file
.............
.............
1999/06/04 13:51:09| comm_write: fd_table[323].rwstate
!= NULL
1999/06/04 13:51:12| comm_write: fd_table[491].rwstate
!= NULL
1999/06/04 13:51:12| comm_write: fd_table[505].rwstate
!= NULL
1999/06/04 13:51:12| comm_write: fd_table[508].rwstate
!= NULL
1999/06/04 13:53:55| comm_write: fd_table[906].rwstate
!= NULL
I have changed the NUMTHREADS 128 in ...src/aiops.c
....
Polygraph server can handle 500 req/sec without proxy
for an hour.
Any help will be greatly appreciated.....
Gyan Prakash..
Squid user...
_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com
Received on Mon Jun 14 1999 - 17:14:42 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:46:51 MST