On 04/15/2011 12:46 PM, Marek Marczykowski wrote: > On 04/14/11 16:15, Raoul Bhatia [IPAX] wrote: >> On 04/13/2011 02:46 PM, Florian Haas wrote: >>>> things i noticed: >>>> >>>> 1. the ra i'm running does not catch "connection refused" errors. >>>> c/p from the replication error: >>>> >>>>> Last_IO_Errno: 2013 >>>>> Last_IO_Error: error connecting to master >>>>> 'mysql_rep@wdb01:3306' - retry-time: 60 retries: 86400 >>>> >>>> i run into this error because of a firewall misconfiguration. >>> >>> Well if the firewall just dropped the packets (as opposed to returning a >>> RST TCP packet), then there wouldn't be any "connection refused" error >>> to be expected. >> >> florian, i don't get what you're trying to tell me. >> this is an error which the ra did not catch. >> the ra reported: up and running (in sync) whereas the slave >> clearly wasn't. >> >> read: the ra behaved wrong with this error. > > Yes, but this is difficult problem: this can also happen, when master > just failed and monitor action doesn't detect it yet. In that case we > don't want to assume that slave is outdated/out of order. > > So current behaviour is to left monitoring this to the user and do not > fail monitor action when slave cannot connect to master (but issue > warning in log message). > > Have you some idea how it could be done to not break working slave, when > the master fails?
is there a possibility to check if pacemaker is happy with the currently promoted master instance? or maybe there should be a retry limit? e.g. if the status does not change for a certain time, an error can be generated? unfortunately, i do not know that much about m/s handling inside a ra... cheers, raoul -- ____________________________________________________________________ DI (FH) Raoul Bhatia M.Sc. email. [email protected] Technischer Leiter IPAX - Aloy Bhatia Hava OG web. http://www.ipax.at Barawitzkagasse 10/2/2/11 email. [email protected] 1190 Wien tel. +43 1 3670030 FN 277995t HG Wien fax. +43 1 3670030 15 ____________________________________________________________________ _______________________________________________________ Linux-HA-Dev: [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev Home Page: http://linux-ha.org/
