Bug #40865 ALL DUMP/REPORT commands return result to all ndb_mgm windows.
Submitted: 19 Nov 2008 21:02 Modified: 3 Mar 2010 12:03
Reporter: Matthew Montgomery Email Updates:
Status: Closed Impact on me:
None 
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S3 (Non-critical)
Version:mysql-5.1-telco-7.0 OS:Any
Assigned to: Magnus Blåudd CPU Architecture:Any
Tags: mysql-5.1-telco-6.4 -> ndb-6.4.0, ndb-6.3.18

[19 Nov 2008 21:02] Matthew Montgomery
Description:
ALL DUMP 1000 and ALL REPORT MEMORY USAGE should behave similarly to ALL STATUS such that the ndb_mgm client which issued the command should be the only one to receive the output results.

Monitoring applications which rely upon ALL REPORT/ALL DUMP signals to track memory usage could flood consoles of administrators trying to operate in the ndb_mgm interface with unwanted messages.

How to repeat:
Setup basic cluster, 1 ndb node is fine.

[NDBD DEFAULT]
NoOfReplicas	= 1
Datadir		= /sandbox/64/cluster
#MemReportFrequency	= 60
MaxNoOfConcurrentOperations=20000

[NDB_MGMD]
Hostname	= localhost
Datadir		= /sandbox/64/cluster

[NDBD]
Hostname	= localhost

(terminal 1)
ndb_mgm> all dump 1000
Sending dump signal with data:
0x000003e8 
ndb_mgm> Node 2: Data usage is 0%(2 32K pages of total 2560)
Node 2: Index usage is 0%(5 8K pages of total 2336)

(terminal 2)
ndb_mgm> 
ndb_mgm> Node 2: Data usage is 0%(2 32K pages of total 2560)
Node 2: Index usage is 0%(5 8K pages of total 2336)

Suggested fix:
Change ndb_mgm to subscribe to that event type once an ALL DUMP is called and then unsubscribe after the result is received.

This would also prevent the MemReportFrequency option from flooding the ndb_mgm consoles with memory usage messages.
[20 Nov 2008 13:21] Jonas Oreland
if it acceptable that the "extra" gets into the cluster log,
it shouldnt be insanely hard to subscribe/dump/unsubscribe
in a "reasonably" sane manner.

ofcourse we really really dont what to muck about with such a things...
R2, E5
[2 Mar 2010 9:29] 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/commits/102005
[2 Mar 2010 17:59] Bugs System
Pushed into 5.1.41-ndb-7.0.14 (revid:magnus.blaudd@sun.com-20100302174920-7j3bpj5kjfg54b69) (version source revid:magnus.blaudd@sun.com-20100302171239-97hayyf3l0q2rrz3) (merge vers: 5.1.41-ndb-7.0.14) (pib:16)
[2 Mar 2010 18:00] Bugs System
Pushed into 5.1.41-ndb-6.3.33 (revid:magnus.blaudd@sun.com-20100302171238-yphjt5qcs8epldzm) (version source revid:magnus.blaudd@sun.com-20100302171238-yphjt5qcs8epldzm) (merge vers: 5.1.41-ndb-6.3.33) (pib:16)
[3 Mar 2010 10:16] Magnus Blåudd
Fixed by patch for bug#31542

Pushed to 6.3.33, 7.0.14 and 7.1.2-beta
[3 Mar 2010 12:03] Jon Stephens
Documented bugfix in the NDB-6.3.33, 7.0.14, and 7.1.2 changelogs, as follows:

        Formerly, the REPORT and DUMP commands returned output to all
        ndb_mgm clients connected to the same MySQL Cluster. Now, these
        commands return their output only to the ndb_mgm client that
        actually issued the command.

Closed.