Hi,
I've been running our squid proxies for a while now. I had mistakenly set it up for diskd on a Linux box. I switched to aufs as various squid list messages stated that's preferable for Linux.
Well when I did that I noticed that VSZ spikes to 175000 or about that as soon as it gets a request. Is this normal? RSS would stay at about 6000. Note this is a empty cache at this point so there isn't anything in the cache_dir.
I switched it to ufs and VSZ & RSS are currently 10308 and 6520. I did a search and couldn't find anything relevent.
I'm using ps -aux to show these numbers incase that matters.
Thanks, Billy
Details:
Fedora Core release 1 Squid 2.5-STABLE6 + ntlm_noreuse_leak patch 2.40 Ghz Xeon 512MB 80G SCSI
Also runs spamassasin/postfix
configure:
./configure --prefix=/usr/local/squid-2.5.STABLE6.patched \
--enable-snmp \
--enable-underscores \
--enable-auth=basic,ntlm \
--enable-external-acl-helpers=wbinfo_group \
--enable-storeio=aufs,ufs,diskd \
--enable-removal-policies=heap,lru
squid.conf without acl's : (note currently ufs but the spike is when aufs)
auth_param ntlm program /usr/local/samba/bin/ntlm_auth --debuglevel=0 --helper-protocol=squid-2.5-ntlmssp
auth_param ntlm children 15
auth_param ntlm max_challenge_reuses 0
auth_param ntlm max_challenge_lifetime 2 minutes
auth_param basic program /usr/local/samba/bin/ntlm_auth --helper-protocol=squid-2.5-basic
auth_param basic children 2
auth_param basic realm Squid proxy-caching web server
auth_param basic credentialsttl 2 hours
http_reply_access allow all
icp_access allow all
logfile_rotate 7
always_direct allow all
http_port 2122
icp_port 3130
snmp_port 3401
hierarchy_stoplist cgi-bin ?
no_cache deny QUERY
cache_replacement_policy heap GDSF
memory_replacement_policy heap GDSF
cache_swap_low 94
cache_swap_high 95
cache_dir ufs /usr/local/squid-2.5.STABLE6.patched/var/cache 500 16 128
cache_store_log none
refresh_pattern ^ftp: 1440 20% 10080
refresh_pattern ^gopher: 1440 0% 1440
refresh_pattern . 0 20% 4320
shutdown_lifetime 5 second
