Details
-
Bug
-
Status: Open (View Workflow)
-
Major
-
Resolution: Unresolved
-
10.2.9
-
None
-
None
-
None
-
Ubuntu, php web app
Description
We have a big database, about 750GB, we have a Master and 4 slaves. some times without any issues, MariaDB threads increasing suddenly from 300~400 to around 5K and its make MariaDB unavailable and web app get `connection time out`.
By chance we found that we can recover MariaDB trough stoping slaves and starting them after 30 second or 1~2 minute.
But this some times work and some times not work ! and in this situations we should stop MariaDB on Master server and staring it again !
its happened after MariaDB 10.2.9, before. when we were at MariaDB 10.1.X everything were fine !!!
PS : I should mention that some times one of the slaves face gap and it cannot sync data with Master ! and we should stop MariaDB server on this slave server and starting it again.
PS : Also we have `Lock process` in MariaDB 10.2.9 !!!!! before we didn't have this much !!!!!
One of the reason that I check when MariaDB threads went up was these `Locks` , I should mention that these locks are not natural and they are not for my application before we didn't have such a issue on MariaDB 10.1.X
