[squid-users] Re: Weird edge condition with slow FTP transfers via Squid 2.5 stable 1

Ha! Got the solution.

I was using aufs on the back end storage. Changed to ufs and it behaves
as expected (very fast).

Sorry to waste your time - but hopefully this message will answer
anyone else who has this problem in the future.

Best wishes

Tim

On Wed, 16 Apr 2003 13:20:44 +0100
Tim Southerwood <ts@doc.ic.ac.uk> wrote:

> Hello folks,
>
> Got a weird edge case with squid to do with slow FTP gets from the cache in the case of one client accessing the file. Speed goes up massively
> when two clients or more want the same cached file.
>
> Scenario:
>
> Squid cache machine is dual Xeon 2GHz, 2GB RAM, 3 SCSI disks
> (2 dedicated to squid cache), 100Mb link (will be 1Gb later).
>
> 1) Test condition: config squid to allow it to cache upto 1GB files.
>
> 2) Use wget to retrieve Mandrake ISO (651MB) via cache from offsite ftp server. Transfer speed is > 1MB/sec (it's a good ftp server!)
>
> 3) Reget the same file from one client in same way. Speed is now a
> feeble (compared to no cache) 220KB/sec.
>
> 4) Here's the weird part. Simultaneously wget same file from same server via the cache on same OR other PC. Speed of BOTH clients rockets to
> 3.6MB/sec sustained average.
>
> Try a 3rd client, speed drops slightly but only because the 100Mb link to the squid cache is saturated. This shows that there is something special about the 1 client case.
>
> Doing the same test over HTTP yields exact the same results implying
> the problem may be to do with the back end storage part of squid???
>
> This is just bugging me a bit because I hate to put in a cache that reduces speed by a significant factor in certain cases.
>
> I'm sure it's a config problem on my part - does anyone have any theories as to what I might be looking for?
>
> (Disclaimer - I've been over the FAQ, google, mailing lists already).
>
> I would be most appreciative of a pointer on this one. The relevant parts (non comments) are appended to the end of this mail should it help...
>
>
> Best wishes,
> Tim Southerwood
>
> *****Squid.conf******
> http_port 3128
> hierarchy_stoplist cgi-bin ?
> acl QUERY urlpath_regex cgi-bin \?
> no_cache deny QUERY
> cache_mem 1600 MB
> cache_swap_low 90
> cache_swap_high 95
> maximum_object_size 1000 MB
> minimum_object_size 0 KB
> maximum_object_size_in_memory 128 MB
> ipcache_size 65536
> ipcache_low 90
> ipcache_high 95
> fqdncache_size 4096
> cache_replacement_policy lru
> memory_replacement_policy lru
> cache_dir aufs /var/squid/cache 128000 16 16
> cache_dir aufs /var/squid/cache1 128000 16 16
> cache_access_log /var/log/squid/access.log
> cache_log /var/log/squid/cache.log
> cache_store_log /var/log/squid/store.log
> log_ip_on_direct on
> mime_table /etc/squid/mime.conf
> log_mime_hdrs off
> pid_filename /var/run/squid.pid
> debug_options ALL,1
> log_fqdn off
> client_netmask 255.255.255.255
> ftp_user squid@doc.ic.ac.uk
> ftp_list_width 50
> ftp_passive off
> ftp_sanitycheck on
> diskd_program /usr/sbin/diskd
> unlinkd_program /usr/sbin/unlinkd
> auth_param basic children 5
> auth_param basic realm Squid proxy-caching web server
> auth_param basic credentialsttl 2 hours
> request_header_max_size 10 KB
> request_body_max_size 0 KB
> refresh_pattern ^ftp: 1440 20% 10080
> refresh_pattern ^gopher: 1440 0% 10080
> refresh_pattern . 0 20% 10080
> refresh_pattern microsoft\.com\/ 0 20% 10080
> refresh_pattern scripps\.com\/ 10880 0% 10080
> quick_abort_min 10 MB
> quick_abort_max 10 MB
> quick_abort_pct 70
> negative_ttl 10 minutes
> positive_dns_ttl 24 hours
> negative_dns_ttl 10 minutes
> connect_timeout 1 minutes
> read_timeout 5 minutes
> request_timeout 2 minutes
> persistent_request_timeout 10 minute
> client_lifetime 1 day
> half_closed_clients on
> pconn_timeout 300 seconds
> acl all src 0.0.0.0/0.0.0.0
> acl manager proto cache_object
> acl localhost src 127.0.0.1/255.255.255.255
> acl to_localhost dst 127.0.0.0/8
> acl SSL_ports port 443 563 873
> acl Safe_ports port 80 # http
> acl Safe_ports port 81 873
> acl Safe_ports port 21 # ftp
> acl Safe_ports port 443 563 # https, snews
> acl Safe_ports port 70 # gopher
> acl Safe_ports port 210 # wais
> acl Safe_ports port 1025-65535 # unregistered ports
> acl Safe_ports port 280 # http-mgmt
> acl Safe_ports port 488 # gss-http
> acl Safe_ports port 591 # filemaker
> acl Safe_ports port 777 # multiling http
> acl CONNECT method CONNECT
> acl docnet src 146.169.0.0/255.255.0.0
> acl icnet src 155.198.17.114/255.255.255.255
> http_access allow manager localhost
> http_access deny manager
> http_access deny !Safe_ports
> http_access deny CONNECT !SSL_ports
> http_access allow localhost
> http_access allow docnet
> http_access allow icnet
> http_access deny all
> http_reply_access allow localhost
> http_reply_access allow docnet
> http_reply_access allow icnet
> http_reply_access deny all
> icp_access allow localhost
> icp_access allow docnet
> icp_access allow icnet
> icp_access deny all
> miss_access allow localhost
> miss_access allow docnet
> miss_access allow icnet
> miss_access deny all
> reply_body_max_size 0 allow all
> cache_mgr ts@doc.ic.ac.uk
> cache_effective_user squid
> cache_effective_group squid
> visible_hostname wwwcachetest.doc.ic.ac.uk
> logfile_rotate 100
> append_domain .doc.ic.ac.uk
> err_html_text http://www.doc.ic.ac.uk/csg/
> memory_pools on
> forwarded_for on
> cachemgr_passwd disable shutdown
> cachemgr_passwd csgcsg info stats/objects
> cachemgr_passwd disable all
> client_db on
> buffered_logs on
> reload_into_ims off
> as_whois_server whois.ra.net
> delay_pools 2 # 2 delay pools
> delay_class 1 2 # pool 1 is a class 2 pool
> delay_class 2 3 # pool 2 is a class 3 pool
> offline_mode off
> uri_whitespace strip
> coredump_dir /var/squid/cache
> high_response_time_warning 100
> high_page_fault_warning 100
> store_dir_select_algorithm round-robin
> *********************
Received on Wed Apr 16 2003 - 07:24:50 MDT