MXOSRVR stay around even when trafodion goes down
Bug #1415187 reported by
Guy Groulx
This bug affects 1 person
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| Trafodion |
Fix Committed
|
High
|
Tharak Capirala | ||
Bug Description
Recovery scenario.
Killed off a regionserver. Recovery kicked-in but for some reason, a watchdog timer poped on a node and the trafodion came down. IE: Trafodion shut down.
However, several mxosrvrs stayed up. Had to explicitly kill them.
From: Govindarajan, Selvaganes
This issue happens due to node down situation. ESPS, ARKCMP, RMS processes all react to the node down/instance down messages by simply exiting. MXOSRVR also need to do exit without for anything to complete. In case of node down situation, there is no order in which processes can go down.
Selva
| Changed in trafodion: | |
| milestone: | none → r1.1 |
| assignee: | nobody → Tharak Capirala (capirala-tharaknath) |
Revision history for this message
| Tharak Capirala (capirala-tharaknath) wrote | #1 |
| Changed in trafodion: | |
| status: | New → Fix Committed |
To post a comment you must log in.
The repository thread was not being shutdown when processing system messages to exit the process. This has been fixed now. Change ID : Ib7b5de187ff02d