Bug #49884 ndb_waiter retries should be documented
Submitted: 22 Dec 2009 19:48
Reporter: Andrew Hutchings Email Updates:
Status: Verified Impact on me:
None 
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S3 (Non-critical)
Version: OS:Any
Assigned to: CPU Architecture:Any
Triage: Triaged: D4 (Minor) / R6 (Needs Assessment) / E6 (Needs Assessment)

[22 Dec 2009 19:48] Andrew Hutchings
Description:
http://dev.mysql.com/doc/refman/5.1/en/mysql-cluster-programs-ndb-waiter.html

The documentation above needs something along the lines of:

If the current state is not NO_CONTACT, NOT_STARTED or SINGLEUSER (STARTING for example) ndb_waiter with retry 10 times the timeout setting displaying the following on each try:

waitNodeState(STARTED, -1) resetting timeout x

(where x is 1 - 10)

How to repeat:
1. Get an ndbd node is a long STARTING state
2. ndb_waiter -c127.0.0.1 -t 4

(also inspect source in storage/ndb/tools/waiter.cpp)
[22 Dec 2009 19:54] Andrew Hutchings
sorry, that should have been x is 0 - 11.  So a total of 12 x timeout
[22 Dec 2009 19:55] Andrew Hutchings
gah!  I mean 0 - 10, 12 x timeout (there is a try before retry 0)
[7 Jan 2010 10:42] Jon Stephens
It makes no sense to have a timeout option that isn't honoured.

I am resistant to documenting a misfeature and thereby enshrining it as a feature.

I suggest that the misbehaviour be fixed instead.

For this reason, I am changing category to Server:Cluster and lead to BOcklin.
[17 Mar 2010 0:16] Jon Stephens
Removing myself as assignee for reasons stated in previous comment.

NB: Bugs assigned to me should be in the Docs category, not Server.