Bug #33965 all ndbds get sigsegv at the same time
Submitted: 21 Jan 2008 19:43 Modified: 19 Jun 2009 12:21
Reporter: Axel Schwenke Email Updates:
Status: No Feedback Impact on me:
None 
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S1 (Critical)
Version:mysql-5.0 OS:Linux (2.6.13/x86_64)
Assigned to: Assigned Account CPU Architecture:Any
Tags: 5.0.46

[21 Jan 2008 19:43] Axel Schwenke
Description:
All 4 ndbd processes in the cluster die of sigsegv at the same time. There is plenty on memory available. Cluster log, error logs and trace logs are available.

How to repeat:
n/a
[21 Jan 2008 22:51] Jonas Oreland
found a *very* large index scan
keywords: 6912
table: 728
attrInfo: 17318

from api 21 to ndbd 11, this seems to find a incorrectly handled
buffer shortage

---

the query is likely a multi-read-range with huge pushdown-filter
select * from A where B in (...) and C in (...)
where B gets index access and C gets interpreted.

---
[21 Jan 2008 22:52] Jonas Oreland
note: scan_tabreq in ndb_11_tracelog.3 r.sigId: -709358728
[19 May 2009 12:21] Jonathan Miller
Is this reproduced on the latest version?
[19 Jun 2009 23:00] Bugs System
No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".