Bug #19766 Starting lots of/to much API apps causes ndb_mgmd
Submitted: 12 May 2006 12:20 Modified: 19 Jun 2006 8:11
Reporter: Geert Vanderkelen Email Updates:
Status: No Feedback Impact on me:
None 
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S2 (Serious)
Version:5.1bk OS:
Assigned to: CPU Architecture:Any

[12 May 2006 12:20] Geert Vanderkelen
Description:

When configuring a cluster for only 16 nodes, where 14 are API nodes, launching 32 of ndbapi (which obviously can't all connect which is ok): it craches the ndb_mgmd.

How to repeat:
NDBAPI uploaded privately: small app doing a connection, and a scan.

[NDB_MGMD DEFAULT]
PortNumber = 1406
Datadir = /data1/users/geert/star

[NDBD DEFAULT]
NoOfReplicas = 1
DataMemory = 1000M
IndexMemory = 200M
Datadir = /data1/users/geert/star

[TCP DEFAULT]
portnumber = 2202

#
# Computer definitions
#

[COMPUTER]
Id = 1
Hostname = ndbsup-priv-1

[COMPUTER]
Id = 2
Hostname = ndbsup-priv-2

[COMPUTER]
Id = 3
Hostname = ndbsup-priv-3

[COMPUTER]
Id = 4
Hostname = ndbsup-priv-4

#
# Nodes definitions
#

# Management nodes
[NDB_MGMD]
Id = 1
ExecuteOnComputer = 2
ArbitrationRank = 1

[MYSQLD]

[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]
[API]

Run the application like:
for((i=0;i<32;i++)); do ./whoops >/dev/null & done

After the first 10 are connected ok, next get ok, and then the ndb_mgmd is gone.
[12 May 2006 12:22] Geert Vanderkelen
ndbapi app

Attachment: whoops.cc (application/octet-stream, text), 2.99 KiB.

[19 May 2006 8:11] Jonas Oreland
Hi, this could potentially be a duplicate of (or the other way around) of
http://bugs.mysql.com/bug.php?id=19930

Could you retest after I pushed that ?
(or test with my patch)
[19 Jun 2006 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".