[SERVER-53] After switching from paired setup to standalone DB should say it is master Created: 18/May/09  Updated: 12/Jul/16  Resolved: 27/May/09

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: None
Fix Version/s: 0.9.3

Type: Bug Priority: Major - P3
Reporter: Michael Dirolf Assignee: Aaron Staple
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Participants:

 Description   

seems to be some weirdness involved after making this switch. not sure exactly what the proper behavior should be though.



 Comments   
Comment by Aaron Staple [ 27/May/09 ]

fixed 8423cdf30f3dc88759fd2959a83541c2bee68e9d

Comment by Aaron Staple [ 27/May/09 ]

Ok, so I think in the failed runs one of the old pair nodes was still up at the time that the other paired node was restarted as standalone. This paired node issued the negotiate command on the node it thought was its peer, but that node wasn't configured to be part of a pair, so it acted as an arbiter and attempted to forward the negotiate command request. Since the standalone node was actually the intended target of the negotiate command, this node forwarded the command to itself and reached a deadlock.

I think I'll fix by doing a dbtemprelease within the negotiate command and also adding a reentrance check in this command.

Comment by Michael Dirolf [ 27/May/09 ]

I think aaron is back on this, trying to reproduce it with whatever i can remember plus help from yulce in IRC

Comment by Aaron Staple [ 27/May/09 ]

I did a simple test – set up a pair, wrote some data, then restarted both nodes as standalone instances and was able to read / write from each. In other words I couldn't reproduce the failure. Mike, could you let me know exactly what you did when you saw this behavior?

Comment by Eliot Horowitz (Inactive) [ 27/May/09 ]

can you do something reasonable with this?

Comment by Aaron Staple [ 19/May/09 ]

Dwight, do you have any thoughts on this?

Comment by Eliot Horowitz (Inactive) [ 18/May/09 ]

Yeah - i'm not sure what - but something.

Maybe a command line argument rather than a command, not sure.

Comment by Aaron Staple [ 18/May/09 ]

Eliot, should manual intervention ( a command ) be required when starting up a db standalone after the same dbpath has been used for a db in paired mode?

Generated at Thu Feb 08 02:52:54 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.