performance decrease after a while

From: Jorg B. <jorg_b@dont-contact.us>
Date: Wed, 26 May 1999 11:29:48 -0700

We are running squid 2.2.STABLE3 as a transparent cache behind a foundry
layer 4 switch... We have noticed that our cache server's performance
decreases after it has been running for about 48 hours. The dedicated disk
space for cached pages is 10 GB which is distributed on 3 Seagate 10000 RPM
scsi drives (using "md" RAID 0). Logging of the access and store files is
turned off to minimize disk I/O. Our cache server is running Linux kernel
2.2.6 (SMP)....

Here is how the hardware looks like:

OS : Linux (Slackware 4beta3) Kernel 2.2.6(smp)
Processors: 2 Pentiums 350's
Memory: 512 MB's
Storage: Software Raid 0 (3 - 10000 RPM seagate U-2 scsi drives) and one
main drive for the OS
Network: 100BaseT FD

Below you can see the current general runtime statistics and the current
config.
As you can see our cache server gets quite busy...

Does anybody know what could cause the slow down after the cache has been
running for a while ?

Thanks for your help...

Jorg B.

------------ runtime statistics ------------
Squid Object Cache: Version 2.2.STABLE3

 Start Time:
            Wed, 26 May 1999 14:09:58 GMT
 Current Time:
            Thu, 27 May 1999 17:16:59 GMT

Connection information for squid:
        Number of clients accessing cache: 695
        Number of HTTP requests received: 1522841
        Number of ICP messages received: 0
        Number of ICP messages sent: 0
        Number of queued ICP replies: 0
        Request failure ratio: 0.00%
        HTTP requests per minute: 936.0
        ICP messages per minute: 0.0
        Select loop called: 16508623 times, 5.913 ms avg
Cache information for squid:
        Request Hit Ratios: 5min: 45.2%, 60min: 42.5%
        Byte Hit Ratios: 5min: 23.7%, 60min: 20.6%
        Storage Swap size: 5881353 KB
        Storage Mem size: 16404 KB
        Storage LRU Expiration Age: 30.00 days
        Mean Object Size: 9.46 KB
        Requests given to unlinkd: 24009
Median Service Times (seconds) 5 min 60 min:
        HTTP Requests (All): 0.07825 0.07825
        Cache Misses: 0.25890 0.23230
        Cache Hits: 0.01309 0.01469
        Near Hits: 0.18699 0.15888
        Not-Modified Replies: 0.01098 0.01235
        DNS Lookups: 0.00997 0.01331
        ICP Queries: 0.00000 0.00000
Resource usage for squid:
        UP Time: 97621.121 seconds
        CPU Time: 15775.110 seconds
        CPU Usage: 16.16%
        CPU Usage, 5 minute avg: 25.04%
        CPU Usage, 60 minute avg: 23.26%
        Maximum Resident Size: 0 KB
        Page faults with physical i/o: 543616
Memory usage for squid via mallinfo():
        Total space in arena: 84710 KB
        Ordinary blocks: 83439 KB 45541 blks
        Small blocks: 0 KB 0 blks
        Holding blocks: 372 KB 2 blks
        Free Small blocks: 0 KB
        Free Ordinary blocks: 1270 KB
        Total in use: 83811 KB 99%
        Total free: 1270 KB 1%
Memory accounted for:
        Total accounted: 67108 KB
File descriptor usage for squid:
        Maximum number of file descriptors: 1024
        Largest file desc currently in use: 364
        Number of file desc currently in use: 321
        Files queued for open: 0
        Available number of file descriptors: 703
        Reserved number of file descriptors: 100
        Store Disk files open: 6
Internal Data Structures:
        622854 StoreEntries
          3183 StoreEntries with MemObjects
          3149 Hot Object Cache Items
        621453 Filemap bits set
        621450 on-disk objects

---------- squid.conf ------------
http_port 3128
icp_port 0
tcp_incoming_address 0.0.0.0
tcp_outgoing_address 255.255.255.255
udp_incoming_address 0.0.0.0
udp_outgoing_address 255.255.255.255
icp_query_timeout 0
mcast_icp_query_timeout 2000
dead_peer_timeout 10 seconds
hierarchy_stoplist cgi-bin
hierarchy_stoplist ?
no_cache Deny QUERY
cache_mem 16777216 bytes
cache_swap_low 90
cache_swap_high 95
maximum_object_size 4194304 bytes
ipcache_size 1024
ipcache_low 90
ipcache_high 95
fqdncache_size 1024
cache_dir /raid0/cache 10000 16 256
cache_access_log /dev/null
cache_log /usr/local/squid/logs/cache.log
cache_store_log none
emulate_httpd_log off
mime_table /usr/local/squid/etc/mime.conf
log_mime_hdrs off
useragent_log none
pid_filename /usr/local/squid/logs/squid.pid
debug_options ALL,1
log_fqdn off
client_netmask 255.255.255.255
ftp_user Squid@
ftp_list_width 32
cache_dns_program /usr/local/squid/bin/dnsserver
dns_children 15
dns_defnames off
dns_nameservers none
unlinkd_program /usr/local/squid/bin/unlinkd
pinger_program /usr/local/squid/bin/pinger
redirect_children 5
redirect_rewrites_host_header on
authenticate_children 5
authenticate_ttl 3600
wais_relay_port 0
request_size 102400 bytes
refresh_pattern ^ftp: 1440 20% 10080

refresh_pattern ^gopher: 1440 0% 1440

refresh_pattern . 0 20% 4320

reference_age 2592000 seconds
quick_abort_min 16 KB
quick_abort_max 16 KB
quick_abort_pct 95
negative_ttl 300 seconds
positive_dns_ttl 21600 seconds
negative_dns_ttl 300 seconds
range_offset_limit 0 bytes
connect_timeout 120 seconds
siteselect_timeout 4 seconds
read_timeout 900 seconds
request_timeout 30 seconds
client_lifetime 86400 seconds
half_closed_clients on
pconn_timeout 120 seconds
ident_timeout 10 seconds
shutdown_lifetime 30 seconds
acl QUERY urlpath_regex cgi-bin
acl QUERY urlpath_regex \?
acl all src 207.173.11.0/255.255.255.192
acl all src 207.173.10.0/255.255.255.0
acl all src 207.173.14.0/255.255.255.0
acl all src 209.63.56.0/255.255.255.0
acl all src 209.63.55.0/255.255.255.0
acl all src 209.63.107.0/255.255.255.0
acl all src 209.210.78.0/255.255.255.0
acl manager proto cache_object
acl localhost src 127.0.0.1
acl webmanage src 209.210.78.22
acl PURGE method PURGE
acl SSL_ports port 443
acl SSL_ports port 563
acl Safe_ports port 80
acl Safe_ports port 443
acl Safe_ports port 563
acl CONNECT method CONNECT
http_access Allow PURGE localhost
http_access Deny PURGE
http_access Allow manager localhost
http_access Allow manager webmanage
http_access Deny manager
http_access Deny !Safe_ports
http_access Deny CONNECT !SSL_ports
http_access Allow all
icp_access Deny all
miss_access Allow all
proxy_auth_realm Squid proxy-caching web server
ident_lookup_access Deny all
cache_mgr support@cwo.com
cache_effective_user squid
cache_effective_group squid
announce_period 31536000 seconds
announce_host tracker.ircache.net
announce_port 3131
httpd_accel_host virtual
httpd_accel_port 80
httpd_accel_with_proxy on
httpd_accel_uses_host_header on
logfile_rotate 10
tcp_recv_bufsize 0 bytes
err_html_text
memory_pools on
memory_pools_limit 0 bytes
forwarded_for on
log_icp_queries on
icp_hit_stale off
minimum_direct_hops 4
cachemgr_passwd XXXXXXXXXX all
store_avg_object_size 13 KB
store_objects_per_bucket 50
client_db on
netdb_low 900
netdb_high 1000
netdb_ping_period 300 seconds
query_icmp off
test_reachability off
buffered_logs off
reload_into_ims off
anonymize_headers
icon_directory /usr/local/squid/etc/icons
error_directory /usr/local/squid/etc/errors
minimum_retry_timeout 5 seconds
maximum_single_addr_tries 3
as_whois_server whois.ra.net
incoming_icp_average 6
incoming_http_average 4
min_icp_poll_cnt 8
min_http_poll_cnt 8
max_open_disk_fds 0
offline_mode off
uri_whitespace deny
prefer_direct on
strip_query_terms on
Received on Wed May 26 1999 - 12:31:59 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:46:27 MST