|
Every time the mcs-controllnode process starts on a non-primary node, it raises an error about the file /var/lib/columnstore/data1/systemFiles/dbrm/tablelocks missing:
$ sudo systemctl restart mcs-controllernode
|
$ sudo systemctl status mcs-controllernode
|
● mcs-controllernode.service - mcs-controllernode
|
Loaded: loaded (/lib/systemd/system/mcs-controllernode.service; static; vendor preset: enabled)
|
Active: active (running) since Mon 2020-08-03 22:15:07 UTC; 1s ago
|
Main PID: 18879 (controllernode)
|
Tasks: 2 (limit: 1074)
|
Memory: 3.2M
|
CGroup: /system.slice/mcs-controllernode.service
|
└─18879 /usr/bin/controllernode fg
|
|
Aug 03 22:15:07 mcs-ubuntu2004-3 systemd[1]: Started mcs-controllernode.
|
Aug 03 22:15:07 mcs-ubuntu2004-3 IDBFile[18879]: 07.994772 |0|0|0| D 35 CAL0002: Failed to open file: /var/lib/columnstore/data1/systemFiles/dbrm/tablelocks, exception: unable to open Buffered file
|
Aug 03 22:15:07 mcs-ubuntu2004-3 controllernode[18879]: 07.995245 |0|0|0| D 29 CAL0000: TableLockServer::load(): could not open the save file/var/lib/columnstore/data1/systemFiles/dbrm/tablelocks
|
I have been told that this is probably "normal" for a non-primary node. If so, then the error message should be improved to clarify that this error is not serious.
If the mcs-controllernode process isn't supposed to run at all on non-primary nodes, then it might be best to avoid starting the process on those nodes in general. Why is ColumnStore starting a process that it doesn't need?
|