Bug #25117 mysql hangs if during creating logfile group or tablespace one ndbd was stopped

• Submitted: 17 Dec 2006 8:19
• Modified: 26 Dec 2006 16:32
• Reporter: Serge Kozlov
• Status: Verified
• Category: MySQL Cluster: Disk Data
• Severity: S3 (Non-critical)
• Version: mysql-5.1
• OS: Linux (Linux FC4)
• Assigned to: Assigned Account
• CPU Architecture: Any
• Tags: 5.1.15-bk

[17 Dec 2006 8:19] Serge Kozlov

Description:
There is a cluster with 2 replicas and 2 data nodes. Mysql hangs if during creating files of logfile group or tablespace the data node was stopped.

How to repeat:
1. Start cluster.

2. Start mysql (client):
 ./mysql -uroot

3. Start ndb_mgm

4. Try to create log file group:

mysql> create logfile group lg1 add undofile 'undofile.dat' initial_size 100M undo_buffer_size 1 engine=ndb;

5. Stop one data node during creating logfile group. 

6. Results are:
 a) In fact undofile.dat is created but not available.
 b) Stopped node couldn't start again
 c) Cluster log contains following records:
2006-12-15 09:14:03 [MgmSrvr] INFO     -- Node 2: DICT: lock bs: 1 ops: 1 poll:
1 cnt: 320 queue: 3
2006-12-15 09:14:10 [MgmSrvr] INFO     -- Node 2: DICT: lock bs: 1 ops: 1 poll:
1 cnt: 384 queue: 3
2006-12-15 09:14:17 [MgmSrvr] INFO     -- Node 2: DICT: lock bs: 1 ops: 1 poll:
1 cnt: 448 queue: 3
2006-12-15 09:14:24 [MgmSrvr] INFO     -- Node 2: DICT: lock bs: 1 ops: 1 poll:
1 cnt: 512 queue: 3
2006-12-15 09:15:19 [MgmSrvr] INFO     -- Node 2: DICT: lock bs: 1 ops: 1 poll:
1 cnt: 1024 queue: 3

[17 Dec 2006 8:53] Serge Kozlov

I stopped non-master node