Bug #41469 Incorrect GCP(save)take-over when GCP_SAVE_REQ has reach all but new master

• Submitted: 15 Dec 2008 12:23
• Modified: 7 Jan 2009 15:21
• Reporter: Jonas Oreland
• Status: Closed
• Category: MySQL Cluster: Cluster (NDB) storage engine
• Severity: S3 (Non-critical)
• Version: *
• OS: Any
• Assigned to: Jonas Oreland
• CPU Architecture: Any

[15 Dec 2008 12:23] Jonas Oreland

Description:
If master dies during GCP_SAVE and the GCP_SAVE_REQ message has reached 
all but next master, then the new master will restart GCP_SAVE with
incorrect GCI value.
(only possible using >1 nodegroup (i.e more than 2-node with 2 replicas)

How to repeat:
run new testprg, with error insert for this specific case
observed by autotest

Suggested fix:
init gcp_save value correctly

[15 Dec 2008 19:35] 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/61701

2779 Jonas Oreland	2008-12-15
      ndb - bug#41469 - incorrectly handling of GCP_SAVE during master-node-failure

[15 Dec 2008 19:41] Bugs System

Pushed into 5.1.30-ndb-6.2.17  (revid:[e-mail]-20081215193537-rnue4wi6wj7etv71) (version source revid:[e-mail]-20081215193537-rnue4wi6wj7etv71) (pib:5)

[15 Dec 2008 19:42] Bugs System

Pushed into 5.1.30-ndb-6.3.20  (revid:[e-mail]-20081215193537-rnue4wi6wj7etv71) (version source revid:[e-mail]-20081215193837-992jjvptoas0f64e) (pib:5)

[15 Dec 2008 19:43] Jonas Oreland

note: 63 version seems wrong

[15 Dec 2008 19:43] Bugs System

Pushed into 5.1.30-ndb-6.4.0  (revid:[e-mail]-20081215193537-rnue4wi6wj7etv71) (version source revid:[e-mail]-20081215194101-sqesh7ogzuenxrjw) (pib:5)

[18 Dec 2008 10:22] Jonas Oreland

already fixed/pushed,
no need to triage

[7 Jan 2009 15:21] Jon Stephens

Documented bug fix in the NDB-6.2.17 and NDB-6.3.21 changelogs as follows:

        If the master node failed during a global checkpoint, it was
        possible in some circumstances for the new master to use an
        incorrect value for the global checkpoint index. This could
        occur only when the cluster used more than one node group.

[10 Feb 2009 20:13] Bugs System

Pushed into 6.0.10-alpha (revid:[e-mail]-20090210194937-s7xshv5l3m1v7wi9) (version source revid:[e-mail]-20090108115759-b4yhuwkm6w8tg7j3) (merge vers: 6.0.10-alpha) (pib:6)

[12 Feb 2009 10:06] 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/66011

2814 He Zhenxing	2009-02-12 [merge]
      Auto merge 6.0 -> 6.0-rpl