[MCOL-3589] Taken down module status doesn't propagate to OAM / show in mcsadmin getsystemstatus Created: 2019-11-05 Updated: 2023-10-26 Resolved: 2019-11-20 |
|
| Status: | Closed |
| Project: | MariaDB ColumnStore |
| Component/s: | ? |
| Affects Version/s: | 1.4.0 |
| Fix Version/s: | 1.4.1 |
| Type: | Bug | Priority: | Critical |
| Reporter: | Jens Röwekamp (Inactive) | Assignee: | Daniel Lee (Inactive) |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | SkySQLMVP | ||
| Environment: |
Multi node ColumnStore 1.4.0 - with and without storage manager Git version engine: 1f47534 |
||
| Issue Links: |
|
||||||||
| Description |
|
In a 1UM xPM setup, the ColumnStore service gets stopped on PM2 through /usr/local/mariadb/columnstore/bin/columnstore stop. How to reproduce:
|
| Comments |
| Comment by Andrew Hutchings (Inactive) [ 2019-11-06 ] |
|
Set to 1.4.1 to investigate |
| Comment by Andrew Hutchings (Inactive) [ 2019-11-12 ] |
|
Confirmed in current 1.4.1 with AWS 2 node combined UM/PM. "columnstore stop" doesn't appear to be doing anything, the node's processes stay up. |
| Comment by Andrew Hutchings (Inactive) [ 2019-11-12 ] |
|
The columnstore script is missing quite a few parts, looking at the history it has been that way since InfiniDB days. It currently just kills ProcMon and ProcMgr on the node and shuts down the local MariaDB, leaving everything else running and nothing to report status. To fix this we need mcsadmin to have ProcMgr's STOPMODULE implemented as a call and then hook it into this script. |
| Comment by Andrew Hutchings (Inactive) [ 2019-11-15 ] |
|
For QA: Ben has added a "stopmodule" command to ColumnStore to stop a specific module. The "columnstore stop" command now uses this. |
| Comment by Daniel Lee (Inactive) [ 2019-11-20 ] |
|
Build verified: 1.4.1-1 Server Engine mcsadmin> getsystemstatus System vagrantTestStack System and Module statuses Component Status Last Status Change Module um1 ACTIVE Wed Nov 20 17:15:29 2019 |