Bug #53934 Incorrect handling of no-logging table during system restart
Submitted: 24 May 2010 7:38 Modified: 25 May 2010 9:08
Reporter: Jonas Oreland Email Updates:
Status: Closed Impact on me:
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S3 (Non-critical)
Version:mysql-5.1-telco-6.2 OS:Any
Assigned to: Jonas Oreland CPU Architecture:Any

[24 May 2010 7:38] Jonas Oreland
During system restart,
a nologging table is not always stored to disk by DIH
This can lead to problems during later system restarts.

How to repeat:
1) start 2-node cluster id = (2,3)
2) stop node 3
3) create nologging table (set ndb_table_no_logging=1)
4) stop node 2
5) start cluster (2,3)
6) ndb_mgm -e "3 error 7221"
7) ndb_mgm -e "2 stop -a"
7) sleep 3
8) ndb_mgm -e "3 stop -a"
9) start cluster
10) crash

Error data: Error opening DIH schema files for table: X


this is sometimes (very rarely) seen in autotest, testDict -n DictRestart

Suggested fix:
Make sure that nologging table's fragment-info files are always written
during system restart.
[24 May 2010 7:39] 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:


3098 Jonas Oreland	2010-05-24
      ndb - bug#53934
        Fix incorrect handling of nologging tables during system-restart
        causing "Error opening DIH schema files for table"
[24 May 2010 7:48] Jonas Oreland
pushed to 6.2.19, 6.3.34, 7.0.15 and 7.1.4
[25 May 2010 9:08] Jon Stephens
Documented bugfix in the NDB-6.2.19, 6.3.34, 7.0.15, and 7.1.4 changelogs, as follows:

      A table created while ndb_table_no_logging was enabled was not 
      always stored to disk, which could lead to a data node crash with 
      -Error opening DIH schema files for table-.