[MCOL-401] systemd entry not created in multi node combined deployment Created: 2016-11-10  Updated: 2016-11-29  Resolved: 2016-11-29

Status: Closed
Project: MariaDB ColumnStore
Component/s: installation
Affects Version/s: 1.0.4
Fix Version/s: 1.0.6

Type: Bug Priority: Major
Reporter: David Thompson (Inactive) Assignee: David Thompson (Inactive)
Resolution: Duplicate Votes: 0
Labels: None
Environment:

CentOS 7


Issue Links:
Relates
relates to MCOL-419 pre-uninstall script is calling sudo ... Closed
Sprint: 2016-22, 2016-23

 Description   

Performed a local 2 node combined install using local disk deployed on 2 seperate vms with static ips on my laptop, also selected rpm install for other node as this is what i installed on first server.

This works great. As a result of testing recovery procedures, uncovered that on the second node, the installer did not install a systemd entry for columnstore. This means that recovery is not automated and the only way to start the system is full shutdown and restart.

After manually creating and enabling for auto start, the system behaves as designed which is that the system auto recovers.



 Comments   
Comment by David Thompson (Inactive) [ 2016-11-18 ]

Ben, could you take a look at this, you probably will need some help from david hill.

Comment by David Thompson (Inactive) [ 2016-11-29 ]

This was caused by MCOL-419 and the secondary system not allowing sudo calls without a tty. Enabling that resolves the problem and the other bug will fully resolve the overall issue.

Comment by David Thompson (Inactive) [ 2016-11-29 ]

Tested with the workaround of commenting out:
Defaults requiretty

in /etc/sudoers in the other nodes and this problem does not occur anymore. Workaround also added to https://mariadb.com/kb/en/mariadb/preparing-for-columnstore-installation/

Generated at Thu Feb 08 02:20:48 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.