[squid-users] Memory leak on epoll support in IA64

This is a discussion on [squid-users] Memory leak on epoll support in IA64 within the Squid Users forums, part of the Web Server and Related Forums category; Hello Adam and all, I am muthukumar working in epoll development on Squid.I am working in IA64 platfrom.Then ...


Go Back   Usenet Forums > Web Server and Related Forums > Squid Users

FAQ Members List Calendar Search Today's Posts Mark Forums Read
  #1 (permalink)  
Old 09-04-2003
MUTHUKUMAR KANDASAMY
 
Posts: n/a
Default [squid-users] Memory leak on epoll support in IA64

Hello Adam and all,

I am muthukumar working in epoll development on Squid.I am working in IA64
platfrom.Then I have tested the squid with epoll for 300 req/sec.But after
that it is consuming more than 1.9 GB out of 2.0 GB.As Adam's advice, I
have changed the swap memory from 1945.634 to 256.000.

[Try shrinking your swap partition in half (to about 1 GB) and see if that
improves your memory situation. You don't have to reboot the
system - just use swapoff to turn off the swap space, use parted to resize
it (check /etc/fstab for the minor partition # first), then use
swapon to turn it back on ]

But still the squid is exiting due to
Sep 4 11:10:19 pandia squid[1184]: Squid Parent: child process 1186 started
Sep 4 11:15:40 pandia kernel: Out of Memory: Killed process 1186 (squid).
Sep 4 11:15:40 pandia squid[1184]: Squid Parent: child process 1186 exited
due to signal 9

I have tuned the kernel parameters of
net.ipv4.ip_forward = 1
net.ipv4.conf.default.rp_filter = 1
kernel.sysrq = 0
kernel.core_uses_pid = 1
fs.file-max = 16384
net.ipv4.ipfrag_low_thresh = 900000
net.ipv4.ipfrag_high_thresh = 1000000
net.ipv4.ipfrag_time = 45
net.ipv4.tcp_rmem = 2000000 2250000 2500000
net.ipv4.tcp_wmem = 1000000 1250000 1500000
net.ipv4.neigh.default.gc_thresh1 = 1024
net.ipv4.neigh.default.gc_thresh2 = 4096
net.ipv4.neigh.default.gc_thresh3 = 8192
net.core.rmem_max = 1500000
net.core.rmem_default = 1500000
net.core.wmem_max = 1000000
net.core.wmem_default = 1000000
I have used the squid configuration of
configure options:
============
'--prefix=/usr/local/squidepoll' '--enable-epoll' '--disable-poll'
'--disable-select' '--disable-kqueue' '--enable-storeio=null,aufs,ufs'
'--with-file-descriptors=16384' '--enable-async-io=16' '--with-pthreads'

I have tried a lot to satisfy more than 300 requests.But i cann't make it
..Any other parameter must be tuned for good performance?.

So please suggest me some information regarding to this.Then regarding to
kernel I am using the sys_epoll enabled 2.4.20 on Ia64.I have made it with
sys_epoll patch and IA64 patch.Using the sys_call in entry.S,i have used the
epoll support on IA64.

I want to know whether this affection is on IA64 kernel or
squid-3.0-pre3.Regarding to the 300 req test also consumed upto 1.8 GB.So
please tell some way to improve the squid for epoll support.The kernel
series 2.5 and 2.6 is supporting sys_epoll.But is their any affection of
kenrel on Squid-3.0-pre.

I have enabled the epoll on 2.4.20 mannually without default on 2.5 and
2.6.So I think ,there is no effection of kernel on Squid-3.0 in the effect
of memory leak.
Anyway to improve the squid-3.0-pre3 for epoll support.

Thanks in advance.

<Muthukumar>
Reply With Quote
Reply


Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are Off
[IMG] code is Off
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On



All times are GMT +1. The time now is 06:14 PM.


Powered by vBulletin® Version 3.6.8
Copyright ©2000 - 2008, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO 3.0.0