On Thu, Dec 13, 2007 at 08:55:15PM +0100, Jan-Michael Th?lken wrote: > > Am 13.12.2007 um 18:18 schrieb Dejan Muhamedagic: > > >Hi, > > > >On Thu, Dec 13, 2007 at 05:31:54PM +0100, Jan-Michael Th?lken wrote: > >> > >>Am 13.12.2007 um 05:34 schrieb Simon Horman: > >> > >>>On Wed, Dec 12, 2007 at 01:38:30PM +0100, Jan-Michael Th?lken wrote: > >>>> > >>>>Am 12.12.2007 um 03:59 schrieb Simon Horman: > >>>> > >>>>>>this works well with lditectord. The Problem occurs only with > >>>>>>mysql :( > >>>>> > >>>>>Can you confirm that mysql will accept connections to the VIP? > >>>>>Perhaps it only accepting connections to the RIP of each real- > >>>>>server, > >>>>>192.168.0.148 and 192.168.0.147 respectively. > >>>> > >>>> > >>>>Loadbalancing is working, that means die VIP accept connections. > >>>>The problem is, that for every negatiote-check a new TIME_WAIT > >>>>appears. > >>> > >>>When you say load balancing is working what do you mean? > >>>Are connections from end-users to VIP:mysql working? > >>>If not it is possible that the real-servers are not > >>>accepting connections made to VIP:mysql. This is because > >>>ldirectord actually checks RIP:mysql not VIP:mysql as checking > >>>the latter is rather tricky to achieve. > >>> > >> > >>all connections are working :) but: > >> > >>every RIP:mysql-check ends in a TIME_WAIT connection as you can see > >>as > >>follow > > > >Haven't followed this closely, but there's nothing wrong with the > >TIME_WAIT state. It's a feature. The TCP puts a connection in > >this state for a while in order not to reuse the same pair until > >the host is reasonably sure that there are not more packets > >originating from this connection wondering through the network. > > >Thanks, > > > >Dejan > > > it's not normal that so many connections are in TIME_WAIT. > and why only mysql connections and not the http and https connections?
AFAIK, all TCP connections should go into the TIME_WAIT state. It's a rule. Of course, this state doesn't last forever, just a few minutes. I'd still maintain that these are nothing to worry about. > and why is the average load so high? You got me now. I really wouldn't know. Thanks, Dejan > > > > > > > > >>lb2:~# netstat > >>Aktive Internetverbindungen (ohne Server) > >>Proto Recv-Q Send-Q Local Address Foreign Address > >>State > >>tcp 0 0 192.168.0.146:54578 192.168.0.148:mysql > >>TIME_WAIT > >>tcp 0 0 192.168.0.146:54590 192.168.0.148:mysql > >>TIME_WAIT > >>tcp 0 0 192.168.0.146:54584 192.168.0.148:mysql > >>TIME_WAIT > >>tcp 0 0 192.168.0.146:54566 192.168.0.148:mysql > >>TIME_WAIT > >>tcp 0 0 192.168.0.146:54572 192.168.0.148:mysql > >>TIME_WAIT > >>tcp 0 0 192.168.0.146:54644 192.168.0.148:mysql > >>TIME_WAIT > >>tcp 0 0 192.168.0.146:54650 192.168.0.148:mysql > >>TIME_WAIT > >>tcp 0 0 192.168.0.146:54626 192.168.0.148:mysql > >>TIME_WAIT > >>tcp 0 0 192.168.0.146:54638 192.168.0.148:mysql > >>TIME_WAIT > >>tcp 0 0 192.168.0.146:54632 192.168.0.148:mysql > >>TIME_WAIT > >> > >> > >>there are about 300 of this connections but only for the mysql not > >>for http > >>or https. > >> > >>every mysql-check leaves a TIME_WAIT connection. > >> > >>if i disable the negotiate check and change it to ping, it works > >>and there > >>are no more TIME_WAITs. > >> > >>the second strange thing on both loadbalancer is, both have a > >>average load > >>of one. the active and the passive > >>and there is no traffic. > >> > >> > >> > >>>-- > >>>Horms > >>> > >>>_______________________________________________ > >>>Linux-HA mailing list > >>>[email protected] > >>>http://lists.linux-ha.org/mailman/listinfo/linux-ha > >>>See also: http://linux-ha.org/ReportingProblems > >> > >>_______________________________________________ > >>Linux-HA mailing list > >>[email protected] > >>http://lists.linux-ha.org/mailman/listinfo/linux-ha > >>See also: http://linux-ha.org/ReportingProblems > >_______________________________________________ > >Linux-HA mailing list > >[email protected] > >http://lists.linux-ha.org/mailman/listinfo/linux-ha > >See also: http://linux-ha.org/ReportingProblems > > _______________________________________________ > Linux-HA mailing list > [email protected] > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems -- Dejan _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
