[Maria-discuss] MariaDB Galera Cluster Wrsep not ready
Hi all, I have a galera cluster of 4 MariaDB server and sometimes a node fail. But when a node fail, others nodes doesn't want to respond to query. I have in log : "MYSQL: query failed: WSREP has not yet prepared node for application use" It's really embarrassing, because it happen like once a week. Is there a parameter to modify to get this working even when split brain ? Like If 2 of the 4 server are ok, I can still make query ? Thanks in advance, -- CHUNKZ.NET - script kiddie and computer technician Bertrand Caplet, Flers (FR) Feel free to send encrypted/signed messages Key ID: 37F70C30 GPG FP: 134A 4027 518B 5F4D D409 558D BA9B 7BF0 37F7 0C30
Hi Bertrand, you should always have an odd number of servers (3, 5, 7) - even number is a risk of split brain, or use weights so you give priority to specific servers. You should also use a load balancer which redirects query to an available node, ex: HAProxy, MaxScale, GLBD... Does the split brain happen because of network issues? 2015-10-24 15:34 GMT+02:00 Bertrand Caplet <bertrand.caplet@chunkz.net>:
Hi all, I have a galera cluster of 4 MariaDB server and sometimes a node fail. But when a node fail, others nodes doesn't want to respond to query. I have in log : "MYSQL: query failed: WSREP has not yet prepared node for application use" It's really embarrassing, because it happen like once a week.
Is there a parameter to modify to get this working even when split brain ? Like If 2 of the 4 server are ok, I can still make query ?
Thanks in advance,
-- CHUNKZ.NET - script kiddie and computer technician Bertrand Caplet, Flers (FR) Feel free to send encrypted/signed messages Key ID: 37F70C30 GPG FP: 134A 4027 518B 5F4D D409 558D BA9B 7BF0 37F7 0C30
_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp
Hi, OK I'll change to odd number. Well I'm planning to set up MaxScale when I get rid of splitbrains. And yes split brain happen cause of network issues. Regards, <guillaume@adishatz.net> wrote :
Hi Bertrand,
you should always have an odd number of servers (3, 5, 7) - even number is a risk of split brain, or use weights so you give priority to specific servers. You should also use a load balancer which redirects query to an available node, ex: HAProxy, MaxScale, GLBD...
Does the split brain happen because of network issues?
-- CHUNKZ.NET - dodgy DIYer and computer technician Bertrand Caplet, Flers (FR) Feel free to send encrypted/signed messages Key ID: FF395BD9 GPG FP: DE10 73FD 17EB 5544 A491 B385 1EDA 35DC FF39 5BD9
participants (2)
-
Bertrand Caplet
-
Guillaume Lefranc