Bug #10669 execute(NoCommit) freeze for ~6seconds
Submitted: 16 May 2005 12:54 Modified: 13 Jun 2005 17:00
Reporter: Johan Andersson Email Updates:
Status: Closed Impact on me:
None 
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S1 (Critical)
Version:4.1++(5.0, 5.1...) OS:Linux (rh9)
Assigned to: Jonas Oreland CPU Architecture:Any

[16 May 2005 12:54] Johan Andersson
Description:
ndbapi problem.
The idea is to check for tuple existence by only doing equal on the primary key (no getValue)
start a transaction
get a committed read operation
equal on the primary key
execute(NoCommit)

The execute hangs for about 6 seconds returning no error

How to repeat:

start a transaction
get a committed read operation
equal on the primary key
execute(NoCommit)

Suggested fix:
-
[16 May 2005 12:56] Johan Andersson
6 seconds should be: 6 minutes
[20 May 2005 1:46] Stewart Smith
Tried to construct a test case, but was unable to trigger the delay described with latest 5.0-ndb tree.

Is it possible to get sample code (real or pseudo)?

This would help in constructing a test case.
[20 May 2005 13:08] Jonas Oreland
I managed to reproduce it...
[20 May 2005 18:02] Johan Andersson
The same problem has been spotted with NdbScanOperation too. That is, scanning on PK only (no getValue)..
[26 May 2005 5:21] 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/internals/25292
[26 May 2005 20:14] Jonas Oreland
I couldn't find any problem with scan.
Is it table or index scan?
[27 May 2005 19:03] 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/internals/25334
[7 Jun 2005 9:54] Jonas Oreland
pushed in 5.0
[9 Jun 2005 5:35] Jonas Oreland
pushed in 4.1.13 and 5.0.7