Bug #40370 master node can die during node-restart

• Submitted: 28 Oct 2008 10:50
• Modified: 29 Oct 2008 6:36
• Reporter: Jonas Oreland
• Status: Closed
• Category: MySQL Cluster: Cluster (NDB) storage engine
• Severity: S3 (Non-critical)
• Version: 6.3.18
• OS: Any
• Assigned to: Jonas Oreland
• CPU Architecture: Any

[28 Oct 2008 10:50] Jonas Oreland

Description:
during node-restart, the master could die with ndbrequire in dbdih
this happened if lcp/gcp was started/triggered exactly during START_ME

this was introduced in 6.3.18 :-( by me

How to repeat:
testNodeRestart -l 10 -n pnr_lcp T1 on >2 node cluster

Suggested fix:
fix incorrect ndbrequire

[28 Oct 2008 10:53] Bugs System

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/57186

2722 Jonas Oreland	2008-10-28
      ndb - bug#40370 - master node can die during node-restart
        - fix incorrect assertion
        - improve testNodeRestart -n pnr to restart half cluster
          instead of just 2 nodes at a time

[28 Oct 2008 10:54] Bugs System

Pushed into 5.1.28-ndb-6.3.19  (revid:[e-mail]-20081028105758-grbcgsv502qapltu) (version source revid:[e-mail]-20081028105758-grbcgsv502qapltu) (pib:5)

[28 Oct 2008 10:58] Bugs System

Pushed into 5.1.28-ndb-6.4.0  (revid:[e-mail]-20081028105758-grbcgsv502qapltu) (version source revid:[e-mail]-20081028110017-9fukws1kacd7vn6b) (pib:5)

[29 Oct 2008 6:36] Jon Stephens

Documented bugfix in the ndb-6.3.19 changelog as follows:

        A restarting data node could fail with an error in the
        DBDIH kernel block if a local or global checkpoint
        was started or triggered just as the node made a request for data from
        another data node.