Re: [squid-users] Squid as Acceleration Proxy in a Squid Cluster
tor 2007-05-10 klockan 10:11 +0200 skrev Marcel Alburg: > But if i stop the local apache on a server and ask over the server the > website - squid sad > > 2007/05/10 08:10:30| Detected DEAD Parent: 127.0.0.1 > 2007/05/10 08:10:30| TCP connection to 127.0.0.1/80 failed > > Thats ok > > but the result is this: > > The following error was encountered: > > * Unable to forward this request at this time. > > 2007/05/10 08:10:55| Failed to select source for 'http://www.xxx.tld' > 2007/05/10 08:10:55| always_direct = 0 > 2007/05/10 08:10:55|never_direct = 0 > 2007/05/10 08:10:55|timedout = 0 For this to work well you need a fallback non-ICP parent relation to the other.. siblings is only queried for the content they have already cached. You should also have a cache_peer_access rule preventing circular forwarding in case both is having problems.. Someting like the following should work: cache_peer the_apache parent ... no-query originserver default cache_peer the_other_server parent ... no-query cache_peer the_other_server sibling ... name=other_server_sibling (note: no no-query on the sibling relation, only the parent relations) acl from_other_server src the_other_server cache_peer_access the_other_server deny from_other_server cache_peer_access the_other_server_sibling deny from_other_server The dual sibling & parent relation to the other server is needed so that ICP only acts in sibling mode, but that it's still used as a parent if the preferred web server is down. Regards Henrik signature.asc Description: Detta är en digitalt signerad meddelandedel
Re: [squid-users] Squid as Acceleration Proxy in a Squid Cluster
My cache peer lines look like this On first server: cache_peer 127.0.0.1parent 80 0 originserverno-query default cache_peer server2sibling 80 3130 proxy-only no-delay allow-miss name=server2 On second server: cache_peer 127.0.0.1parent 80 0 originserverno-query default cache_peer server1sibling 80 3130 proxy-only no-delay allow-miss name=server1 Marcel Alexandre Correa wrote: > add no-query to cache_peer line ! > > i think this 'solve' ... > > On 5/10/07, Marcel Alburg <[EMAIL PROTECTED]> wrote: >> Ok, thanks - that works and i see that a sibling becomes some request >> for caches. >> >> But if i stop the local apache on a server and ask over the server the >> website - squid sad >> >> 2007/05/10 08:10:30| Detected DEAD Parent: 127.0.0.1 >> 2007/05/10 08:10:30| TCP connection to 127.0.0.1/80 failed >> >> Thats ok >> >> but the result is this: >> >> The following error was encountered: >> >> * Unable to forward this request at this time. >> >> 2007/05/10 08:10:55| Failed to select source for 'http://www.domain.tld' >> 2007/05/10 08:10:55| always_direct = 0 >> 2007/05/10 08:10:55|never_direct = 0 >> 2007/05/10 08:10:55|timedout = 0 >> >> >> >> Thanks >> >> Marcel >> >> leongmzlist wrote: >> > 1. your cache sibling configuration is wrong; you're not contacting any >> > of the sibling. >> > 2. use the proxy-only flag if you don't want squid to store cache hits >> > from the siblings. >> > >> > try something like >> > >> > cache_peer sibling1 sibling 80 3130 proxy-only no-delay allow-miss >> > name=sibling1 >> > >> > mike >> > >> > At 03:02 PM 5/9/2007, Marcel Alburg wrote: >> >> Hello, >> >> >> >> my Networks looks like >> >> >> >> >> >> Internet >> >> | >> >> [LB] -ha- [LB] >> >> | | >> >> +-+ >> >> | | >> >> | | >> >> [Squid + Apache] [Squid + Apache] >> >> >> >> >> >> All Hosts has running a webserver at 127.0.0.1:80 and a squid at the >> >> official IP Adresse on Port 80. >> >> >> >> The LoadBalancer ([LB]) forward the request to on of the server. The >> >> Squid checks his Cache and it the cache valid he send from cache >> >> otherwise he ask the local apache. >> >> >> >> We've a lot of hosts with this configuration and we've a problem, that >> >> any squid cache his own version from a webpage (n-Nodes could be n >> >> different Versions from a webpage - because the access and >> >> generationtime differs). >> >> >> >> My question is. It is possible to let ask the Squid the other nodes in >> >> the network if the requested page cached. And has any node a cached >> >> version - than deliver this cached version - otherwise ask the local >> >> apache. >> >> >> >> I tried a lot of different configs but nothing works to my >> contentment. >> >> >> >> a other problem is: >> >> >> >> If ask the squid the local page and gets an Error 404 or 500 he should >> >> ask the other Proxies for any valid cached version and deliver that >> >> version. The Webuser should not see the 500 Errors (if we had a lot of >> >> load at the server, the error 500 pages comes often). >> >> >> >> And i wanna stop the local apache at one node and the squid should >> >> notice that and ask the other nodes. >> >> >> >> One of my not working config was like this >> >> >> >> cache_peer 127.0.0.1parent 80 0 originserverno-query >> default >> >> allow-miss >> >> cache_peer 127.0.0.1sibling 80 3130 >> >> cache_peer 127.0.0.1sibling 80 3130 >> >> cache_peer 127.0.0.1sibling 80 3130 >> >> icp_hit_stale on >> >> >> >> >> >> It is possible to realized that ? >> >> >> >> Thanks >> >> >> >> Marcel >> > >> > >> >> > >
Re: [squid-users] Squid as Acceleration Proxy in a Squid Cluster
My cache peer lines look like this On first server: cache_peer 127.0.0.1parent 80 0 originserverno-query default cache_peer server2sibling 80 3130 proxy-only no-delay allow-miss name=server2 On second server: cache_peer 127.0.0.1parent 80 0 originserverno-query default cache_peer server1sibling 80 3130 proxy-only no-delay allow-miss name=server1 Marcel Alexandre Correa wrote: > add no-query to cache_peer line ! > > i think this 'solve' ... > > On 5/10/07, Marcel Alburg <[EMAIL PROTECTED]> wrote: >> Ok, thanks - that works and i see that a sibling becomes some request >> for caches. >> >> But if i stop the local apache on a server and ask over the server the >> website - squid sad >> >> 2007/05/10 08:10:30| Detected DEAD Parent: 127.0.0.1 >> 2007/05/10 08:10:30| TCP connection to 127.0.0.1/80 failed >> >> Thats ok >> >> but the result is this: >> >> The following error was encountered: >> >> * Unable to forward this request at this time. >> >> 2007/05/10 08:10:55| Failed to select source for 'http://www.xxx.tld' >> 2007/05/10 08:10:55| always_direct = 0 >> 2007/05/10 08:10:55|never_direct = 0 >> 2007/05/10 08:10:55|timedout = 0 >> >> >> >> Thanks >> >> Marcel >> >> leongmzlist wrote: >> > 1. your cache sibling configuration is wrong; you're not contacting any >> > of the sibling. >> > 2. use the proxy-only flag if you don't want squid to store cache hits >> > from the siblings. >> > >> > try something like >> > >> > cache_peer sibling1 sibling 80 3130 proxy-only no-delay allow-miss >> > name=sibling1 >> > >> > mike >> > >> > At 03:02 PM 5/9/2007, Marcel Alburg wrote: >> >> Hello, >> >> >> >> my Networks looks like >> >> >> >> >> >> Internet >> >> | >> >> [LB] -ha- [LB] >> >> | | >> >> +-+ >> >> | | >> >> | | >> >> [Squid + Apache] [Squid + Apache] >> >> >> >> >> >> All Hosts has running a webserver at 127.0.0.1:80 and a squid at the >> >> official IP Adresse on Port 80. >> >> >> >> The LoadBalancer ([LB]) forward the request to on of the server. The >> >> Squid checks his Cache and it the cache valid he send from cache >> >> otherwise he ask the local apache. >> >> >> >> We've a lot of hosts with this configuration and we've a problem, that >> >> any squid cache his own version from a webpage (n-Nodes could be n >> >> different Versions from a webpage - because the access and >> >> generationtime differs). >> >> >> >> My question is. It is possible to let ask the Squid the other nodes in >> >> the network if the requested page cached. And has any node a cached >> >> version - than deliver this cached version - otherwise ask the local >> >> apache. >> >> >> >> I tried a lot of different configs but nothing works to my >> contentment. >> >> >> >> a other problem is: >> >> >> >> If ask the squid the local page and gets an Error 404 or 500 he should >> >> ask the other Proxies for any valid cached version and deliver that >> >> version. The Webuser should not see the 500 Errors (if we had a lot of >> >> load at the server, the error 500 pages comes often). >> >> >> >> And i wanna stop the local apache at one node and the squid should >> >> notice that and ask the other nodes. >> >> >> >> One of my not working config was like this >> >> >> >> cache_peer 127.0.0.1parent 80 0 originserverno-query >> default >> >> allow-miss >> >> cache_peer 127.0.0.1sibling 80 3130 >> >> cache_peer 127.0.0.1sibling 80 3130 >> >> cache_peer 127.0.0.1sibling 80 3130 >> >> icp_hit_stale on >> >> >> >> >> >> It is possible to realized that ? >> >> >> >> Thanks >> >> >> >> Marcel >> > >> > >> >> > >
Re: [squid-users] Squid as Acceleration Proxy in a Squid Cluster
add no-query to cache_peer line ! i think this 'solve' ... On 5/10/07, Marcel Alburg <[EMAIL PROTECTED]> wrote: Ok, thanks - that works and i see that a sibling becomes some request for caches. But if i stop the local apache on a server and ask over the server the website - squid sad 2007/05/10 08:10:30| Detected DEAD Parent: 127.0.0.1 2007/05/10 08:10:30| TCP connection to 127.0.0.1/80 failed Thats ok but the result is this: The following error was encountered: * Unable to forward this request at this time. 2007/05/10 08:10:55| Failed to select source for 'http://www.xxx.tld' 2007/05/10 08:10:55| always_direct = 0 2007/05/10 08:10:55|never_direct = 0 2007/05/10 08:10:55|timedout = 0 Thanks Marcel leongmzlist wrote: > 1. your cache sibling configuration is wrong; you're not contacting any > of the sibling. > 2. use the proxy-only flag if you don't want squid to store cache hits > from the siblings. > > try something like > > cache_peer sibling1 sibling 80 3130 proxy-only no-delay allow-miss > name=sibling1 > > mike > > At 03:02 PM 5/9/2007, Marcel Alburg wrote: >> Hello, >> >> my Networks looks like >> >> >> Internet >> | >> [LB] -ha- [LB] >> | | >> +-+ >> | | >> | | >> [Squid + Apache] [Squid + Apache] >> >> >> All Hosts has running a webserver at 127.0.0.1:80 and a squid at the >> official IP Adresse on Port 80. >> >> The LoadBalancer ([LB]) forward the request to on of the server. The >> Squid checks his Cache and it the cache valid he send from cache >> otherwise he ask the local apache. >> >> We've a lot of hosts with this configuration and we've a problem, that >> any squid cache his own version from a webpage (n-Nodes could be n >> different Versions from a webpage - because the access and >> generationtime differs). >> >> My question is. It is possible to let ask the Squid the other nodes in >> the network if the requested page cached. And has any node a cached >> version - than deliver this cached version - otherwise ask the local >> apache. >> >> I tried a lot of different configs but nothing works to my contentment. >> >> a other problem is: >> >> If ask the squid the local page and gets an Error 404 or 500 he should >> ask the other Proxies for any valid cached version and deliver that >> version. The Webuser should not see the 500 Errors (if we had a lot of >> load at the server, the error 500 pages comes often). >> >> And i wanna stop the local apache at one node and the squid should >> notice that and ask the other nodes. >> >> One of my not working config was like this >> >> cache_peer 127.0.0.1parent 80 0 originserverno-query default >> allow-miss >> cache_peer 127.0.0.1sibling 80 3130 >> cache_peer 127.0.0.1sibling 80 3130 >> cache_peer 127.0.0.1sibling 80 3130 >> icp_hit_stale on >> >> >> It is possible to realized that ? >> >> Thanks >> >> Marcel > > -- Sds. Alexandre J. Correa Onda Internet / OPinguim.net http://www.ondainternet.com.br http://www.opinguim.net
Re: [squid-users] Squid as Acceleration Proxy in a Squid Cluster
Ok, thanks - that works and i see that a sibling becomes some request for caches. But if i stop the local apache on a server and ask over the server the website - squid sad 2007/05/10 08:10:30| Detected DEAD Parent: 127.0.0.1 2007/05/10 08:10:30| TCP connection to 127.0.0.1/80 failed Thats ok but the result is this: The following error was encountered: * Unable to forward this request at this time. 2007/05/10 08:10:55| Failed to select source for 'http://www.xxx.tld' 2007/05/10 08:10:55| always_direct = 0 2007/05/10 08:10:55|never_direct = 0 2007/05/10 08:10:55|timedout = 0 Thanks Marcel leongmzlist wrote: > 1. your cache sibling configuration is wrong; you're not contacting any > of the sibling. > 2. use the proxy-only flag if you don't want squid to store cache hits > from the siblings. > > try something like > > cache_peer sibling1 sibling 80 3130 proxy-only no-delay allow-miss > name=sibling1 > > mike > > At 03:02 PM 5/9/2007, Marcel Alburg wrote: >> Hello, >> >> my Networks looks like >> >> >> Internet >> | >> [LB] -ha- [LB] >> | | >> +-+ >> | | >> | | >> [Squid + Apache] [Squid + Apache] >> >> >> All Hosts has running a webserver at 127.0.0.1:80 and a squid at the >> official IP Adresse on Port 80. >> >> The LoadBalancer ([LB]) forward the request to on of the server. The >> Squid checks his Cache and it the cache valid he send from cache >> otherwise he ask the local apache. >> >> We've a lot of hosts with this configuration and we've a problem, that >> any squid cache his own version from a webpage (n-Nodes could be n >> different Versions from a webpage - because the access and >> generationtime differs). >> >> My question is. It is possible to let ask the Squid the other nodes in >> the network if the requested page cached. And has any node a cached >> version - than deliver this cached version - otherwise ask the local >> apache. >> >> I tried a lot of different configs but nothing works to my contentment. >> >> a other problem is: >> >> If ask the squid the local page and gets an Error 404 or 500 he should >> ask the other Proxies for any valid cached version and deliver that >> version. The Webuser should not see the 500 Errors (if we had a lot of >> load at the server, the error 500 pages comes often). >> >> And i wanna stop the local apache at one node and the squid should >> notice that and ask the other nodes. >> >> One of my not working config was like this >> >> cache_peer 127.0.0.1parent 80 0 originserverno-query default >> allow-miss >> cache_peer 127.0.0.1sibling 80 3130 >> cache_peer 127.0.0.1sibling 80 3130 >> cache_peer 127.0.0.1sibling 80 3130 >> icp_hit_stale on >> >> >> It is possible to realized that ? >> >> Thanks >> >> Marcel > >
Re: [squid-users] Squid as Acceleration Proxy in a Squid Cluster
1. your cache sibling configuration is wrong; you're not contacting any of the sibling. 2. use the proxy-only flag if you don't want squid to store cache hits from the siblings. try something like cache_peer sibling1 sibling 80 3130 proxy-only no-delay allow-miss name=sibling1 mike At 03:02 PM 5/9/2007, Marcel Alburg wrote: Hello, my Networks looks like Internet | [LB] -ha- [LB] | | +-+ | | | | [Squid + Apache] [Squid + Apache] All Hosts has running a webserver at 127.0.0.1:80 and a squid at the official IP Adresse on Port 80. The LoadBalancer ([LB]) forward the request to on of the server. The Squid checks his Cache and it the cache valid he send from cache otherwise he ask the local apache. We've a lot of hosts with this configuration and we've a problem, that any squid cache his own version from a webpage (n-Nodes could be n different Versions from a webpage - because the access and generationtime differs). My question is. It is possible to let ask the Squid the other nodes in the network if the requested page cached. And has any node a cached version - than deliver this cached version - otherwise ask the local apache. I tried a lot of different configs but nothing works to my contentment. a other problem is: If ask the squid the local page and gets an Error 404 or 500 he should ask the other Proxies for any valid cached version and deliver that version. The Webuser should not see the 500 Errors (if we had a lot of load at the server, the error 500 pages comes often). And i wanna stop the local apache at one node and the squid should notice that and ask the other nodes. One of my not working config was like this cache_peer 127.0.0.1parent 80 0 originserverno-query default allow-miss cache_peer 127.0.0.1sibling 80 3130 cache_peer 127.0.0.1sibling 80 3130 cache_peer 127.0.0.1sibling 80 3130 icp_hit_stale on It is possible to realized that ? Thanks Marcel
[squid-users] Squid as Acceleration Proxy in a Squid Cluster
Hello, my Networks looks like Internet | [LB] -ha- [LB] | | +-+ | | | | [Squid + Apache] [Squid + Apache] All Hosts has running a webserver at 127.0.0.1:80 and a squid at the official IP Adresse on Port 80. The LoadBalancer ([LB]) forward the request to on of the server. The Squid checks his Cache and it the cache valid he send from cache otherwise he ask the local apache. We've a lot of hosts with this configuration and we've a problem, that any squid cache his own version from a webpage (n-Nodes could be n different Versions from a webpage - because the access and generationtime differs). My question is. It is possible to let ask the Squid the other nodes in the network if the requested page cached. And has any node a cached version - than deliver this cached version - otherwise ask the local apache. I tried a lot of different configs but nothing works to my contentment. a other problem is: If ask the squid the local page and gets an Error 404 or 500 he should ask the other Proxies for any valid cached version and deliver that version. The Webuser should not see the 500 Errors (if we had a lot of load at the server, the error 500 pages comes often). And i wanna stop the local apache at one node and the squid should notice that and ask the other nodes. One of my not working config was like this cache_peer 127.0.0.1parent 80 0 originserverno-query default allow-miss cache_peer 127.0.0.1sibling 80 3130 cache_peer 127.0.0.1sibling 80 3130 cache_peer 127.0.0.1sibling 80 3130 icp_hit_stale on It is possible to realized that ? Thanks Marcel