Not the answer you need?
Register and ask your own question!

Weird TokuDB deadlocks

azilberazilber EntrantInactive User Role Beginner
Hi all,

We are running an old version of TokuDB (7.5.1) based on MariaDB 5.5 that we haven't upgraded yet. We have a decent amount of data in there on a server with SSD raid-10 drives, and 260GB of ram (170GB used by TokuDB). The binary-logs are on separate raid-10 15k drives. Things have been running quite well for a long time, but lately we've been having weird stalls on simple queries that happen once a day at a random time. The stalls happen during non-peak times, and simple queries, that access a single key get deadlocked and stalled. It stalls for around 2-3 minutes, causing a bit of havoc, but then recovers. sar shows nothing out of the ordinary in terms of swap, i/o, cpu, network or ram usage. io utilization is negligible. I'm thinking that maybe the indexes need to be rebuilt, as the db is a number of years old, but beyond that, we're stumped.

Comments

  • azilberazilber Entrant Inactive User Role Beginner
    I would like to add that we're also seeing inserts blocking updates, and updates blocking inserts, which is very strange. We've also discovered that updates of the type '`x`=`x`+1' seem to be the main culprit...
Sign In or Register to comment.

MySQL, InnoDB, MariaDB and MongoDB are trademarks of their respective owners.
Copyright ©2005 - 2020 Percona LLC. All rights reserved.