Bug #77433 GET_TABINFOREQ dropped causing LCPs to stop and LCP watchdog to fire
Submitted: 22 Jun 2015 14:00 Modified: 25 Jun 2015 7:30
Reporter: Mikael Ronström Email Updates:
Status: Closed Impact on me:
None 
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S2 (Serious)
Version:7.4.7 OS:Any
Assigned to: CPU Architecture:Any

[22 Jun 2015 14:00] Mikael Ronström 
Description:
When calling GET_TABINFOREQ one can queue requests. When dequeueing it seems that the request
can sometimes silently disappear.

How to repeat:
Run autotest daily-basic on tyr13

Suggested fix:
It looks as if it is assembleFragments that returns false after being called with EXECUTE_DIRECT.
Ensure somehow that signal is properly initialised to handle assembleFragments at all times.
[25 Jun 2015 7:30] Jon Stephens 
Documented fix in the NDB 7.1.36, 7.2.21, 7.3.10, and 7.4.7 changelogs, as follows:

    In some cases, the DBDICT block failed to handle repeated
    GET_TABINFOREQ signals after the first one, leading to possible
    node failures and restarts. This could be observed after setting
    a suffificiently high value for MaxNoOfExecutionThreads and low
    value for LcpScanProgressTimeout.
      
Closed.