Skip to:

  1. Skip to Jira Navigation
  2. Skip to Side Navigation
  3. Skip to Main Content

Galera node stuck in MDL conflict and halted cluster writes

Description

Setup

  • 5 node cluster

  • Version -
    percona xtradb server packge - 5.7.31-31.45-3.bionic

mysqld  Ver 5.7.31-34-57 for debian-linux-gnu on x86_64 (Percona XtraDB Cluster (GPL), Release rel34, Revision 7359e4f, WSREP version 31.45, wsrep_31.45)

  • OS - Ubuntu 18.04

  • One node accepts writes and any DDL operations

Order of events

  • A table ALTER was triggered on the writer node using pt-osc.

  • A reader node in the cluster started logging conflicts between the DELETE Trigger being created as a result of pt-osc and SELECTs accessing that table.

  • Very specific events were logged but after a while just the MDL conflict messages

  • This triggered flow control on the cluster and all write activity halted.

  • Stopping mysql process on the reader node restored the cluster.

Logs

Reader node with conflict

Explicit Logs

 

Environment

None

Smart Checklist

Activity

Show:

Aaditya Dubey December 10, 2023 at 7:26 AM

Hi ,

We still haven't heard any news from you. So I assume issue does not persist anymore and will close the ticket. If you disagree just reply and create a follow-up new Jira Ticket.

Lalit Choudhary November 25, 2021 at 1:31 PM

Hi

Thank you for the report.

Please upgrade to the latest PXC version 5.7.35, as there are fixes for such issue deadlock, MDL locks issue fixed in-between versions.

Like:

https://jira.percona.com/browse/PXC-3418

 

Done

Details

Assignee

Reporter

Affects versions

Priority

Smart Checklist

Created October 20, 2021 at 8:59 PM
Updated December 10, 2023 at 7:26 AM
Resolved December 10, 2023 at 7:26 AM
Configure