WSREP: BF lock wait long

Hi everyone, even though this was closed as incomplete, commenting here on our current status with some good news: since 5.6.47 came out, after checking the changelog I have identified that #2684 (fixed in 5.6.47) closely matches our use-case (heavy use of stored procedures, with occasional updates), so we updated quickly, and we allowed full load (up until now on the updated cluster node hanging the most often we allowed only 1 connection) for it to be just like the other members (all of them running 5.6.36, not having BF lock wait long issues), and we did not have BF lock wait issues since then (with 8 days uptime so far, up until now we had cluster hang once a week, or in some cases even twice a day).

Hello @Michal Hornych,
It's been 52 days since this issue went into Incomplete and we haven't heard
from you on this.

At this point, our policy is to Close this issue, to keep things from getting
too cluttered. If you have more information about this issue and wish to
reopen it, please reply with a comment containing "jira-bot=reopen".

Hi,
let me recap our situation regarding the error with "BF lock wait long".

We installed and started using Pecona Cluster for production environment in March 2019. We are using a cluster consisting of 3 nodes of the same HW configuration. At that time, we used all 3 nodes for write (write-write-write configuration).
Since there was often a "BF lock wait long" error, we tried to reconfigure the proxy, so SQL writes were primarily performed by only one node, while the others were used for reads (write-read-read configuration).

In this configuration, the stability was improved significantly, the "BF lock wait long" error disappears. But there was an another error, freezing the entire server (see my comment from 21.Lis 2019). It was also fixed (big thanks for Krunal Bauskar). The error was caused by setting timezone after each connection established - we disabled it.

Now, the server is running smoothly, but still in write-read-read configuration. It is possible that frequent timezone settings also caused a "BF lock wait long" problem, but this cannot be easily confirmed.
We are continuously installing updates, it is possible that some of them have solved the error, but unfortunately, we cannot confirm it too.

We are considering trying to use the cluster in write-write-write mode, where all nodes will run write queries.

if the problem will occur again in the future, I'll write again
thanks to everyone for the comments.

Jan

Hello @Michal Hornych,
It's jira-bot again. Your bug report is important to us, but we haven't heard
from you since the previous notification. If we don't hear from you on
this in 7 days, the issue will be automatically closed.

Hello @Michal Hornych,
I'm jira-bot, Percona's automated helper script. Your bug report is important
to us but we've been unable to reproduce it, and asked you for more
information. If we haven't heard from you on this in 3 more weeks, the issue
will be automatically closed.