Poll: Which of the following stability problems have impacted you when using MariaDB in the last year?
Our third poll aims to get to know you better by learning which of the following stability problems have impacted you when using MariaDB in the last year. This helps us narrow down the main issues in MariaDB Server’s stability and helps us prioritize accordingly to solve them.
Respondents can vote on each point, allocating between 1 and 4 stars:
- 1 stars – not affected
- 2 star – low impact
- 3 star – high impact
- 4 stars – critical impact
Hello,
At this moment i use MariaDB with no problems else one thing witch is using blob field in photos storing for early versions on windows 7.
Thank you
Having serious memory leaks since upgrading MariaDb from 5.5 to 10.5 (through every major version in between). Now mariadb process leaks tens of GBs of RAM over the course of few months, which requires restarting it, once it about to leak all the available RAM in the system. 5.5 was rock-solid and worked for years without having to restart. Really hope it will be addressed in the upcoming update.
Exactly the same to us.
My biggest rubs I’ve been running into since adopting 10.5.x in the last year have been:
1) Async Parallel Row replication with very high I/O rate’s and periodic replication failures resulting in broken cluster ~ 10.5.8 (might be solved already, we’ve got it disabled for now)
2) Query Optimizer Bug that causes MariaDB to choose the wrong index leading to very long queries ~ 10.5.8 – 10.5.12 (ticket is open, and there is a workaround in place, not sure it’s fixed yet)
3) In Place Algorithm for Optimize Table apparently working, but resulting in a very broken system ~ 10.5.12 (just discovered yesterday. Results in slower queries and growing table space as opposed to table space reuse, table space was 85GB, optimized should have been about 15GB remaining, a day later, table space is 145GB, no ticket yet)
4) Table Definition Cache Value of -1 leading to OOM database crashes. System creates many many temp tables and persistent connections are used, might be related ~ 10.5.8 – 10.5.12 (ticket is open, and yea, I’m behind on follow-up)
* Undocumented of not following least surprise changes without proper justification, sometimes breaking replication
* Memory leaks forcing to restart the server every X weeks
* GTID model not working for a dynamic environment and backups recovery
* Frequent crashes due to data corruption of indexes
* mysql_upgrade not updating grants in an expected way
1 Still not solved https://jira.mariadb.org/browse/MDEV-11588
2 Disgusting implementation of JSON, instead of just doing it as in MySQL.
download 10.2.25 and orthers
We have problem with galera 4.8/4.9 and mariadb 10.5.12 with synced->donor->joined states: https://github.com/codership/galera/commit/065e484144c5999709ae8fd19844da72bb785073#diff-1420386eacc5a85a62fbce99764f1ea7c94cd92af199f62c8a0057b44718e1d1R2021
also facing this issue with v10.5.12: https://jira.mariadb.org/browse/MDEV-26298