[MCOL-3985] multinode Non-systemd installation with replication enabled fails to start Created: 2020-05-07 Updated: 2020-05-14 Resolved: 2020-05-14 |
|
| Status: | Closed |
| Project: | MariaDB ColumnStore |
| Component/s: | installation |
| Affects Version/s: | 1.4.0 |
| Fix Version/s: | 1.4.4 |
| Type: | Bug | Priority: | Major |
| Reporter: | Ben Thompson (Inactive) | Assignee: | Ben Thompson (Inactive) |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Environment: |
Multinode nonsystemd installation with replication |
||
| Description |
|
Mulitnode configuration with replication and no systemd. Installation fails on replication setup because it attempts to restart mysqld process and succeeds in killing mysqld but tries to start new mysqld process before previous pid is gone. the system results in no mysqld process running and no notification in mcsadmin because it never acknowledge the pid changed or disappears. To reproduce build server with systemd features disabled and install columnstore on multinode with replication enabled |
| Comments |
| Comment by Patrick LeBlanc (Inactive) [ 2020-05-08 ] |
|
this needs to go into 1.5 as well, reopening |
| Comment by Patrick LeBlanc (Inactive) [ 2020-05-11 ] |
|
I think I've heard "1.5 is a no-go w/o the OAM replacement" enough times to think this doesn't need to go into 1.5. |
| Comment by Daniel Lee (Inactive) [ 2020-05-13 ] |
|
Please provide step-by-step test scenario on reproducing this the issue and how to verify the fix. thx. |
| Comment by Daniel Lee (Inactive) [ 2020-05-14 ] |
|
Build verified: 1.4.4-1 special build from Patrick This special release package was built with systemd disabled. Installed and tested a 3pm glusterfs stack. |
