Bug #4096 WatchDog expires when thrashing in ndb start
Submitted: 10 Jun 2004 20:17 Modified: 10 Aug 2004 17:13
Reporter: Mikael Ronström Email Updates:
Status: Closed Impact on me:
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S3 (Non-critical)
Version:4.1.3 OS:Mac OS X (Mac OS X)
Assigned to: Mikael Ronström CPU Architecture:Any

[10 Jun 2004 20:17] Mikael Ronström
If the machine is thrashing and the size of the Data Memory is large then it is common
that the WatchDog expires in the start phase 1 while initialising the pages

How to repeat:
Start a bigger system than your memory can handle and a fairly large DataMemory.
E.g. use mysql-test-run --with-ndbcluster on laptop

Suggested fix:
Refresh watchdog in loop while initialising pages
[14 Jun 2004 17:52] Jonas Oreland
Does this happen only in debug mode? (or in release aswell)
In debug there's a memset(someValue) on each piece of memory allocated.
This will make likelyhood of watchdog failure greater, maybe we should remove that.

Otherwise, maybe we should try to timeslice malloc/init instead of disabling watchdog.