[MCOL-5158] MCOL-4939 Causes Regression With Docker Clusters Created: 2022-07-14  Updated: 2022-07-15  Resolved: 2022-07-15

Status: Closed
Project: MariaDB ColumnStore
Component/s: cmapi
Affects Version/s: cmapi-6.4.1
Fix Version/s: N/A

Type: Bug Priority: Blocker
Reporter: Todd Stoffel (Inactive) Assignee: Alan Mologorsky
Resolution: Not a Bug Votes: 0
Labels: None

Issue Links:
Problem/Incident
is caused by MCOL-4939 Add a method to disable failover faci... Closed
Relates
relates to MCOL-4218 Remove NFS/EFS/ Filestore prereq for... Stalled
relates to MCOL-5157 Remove Unnecessary Message about Fail... Closed
Sprint: 2021-17

 Description   

Whatever changes were implemented in MCOL-4939 included some logic to determine whether or not shared storage is being used. This check is inadequate as it fails to recognize shared volumes within a Docker environment and this will break our publicly facing Docker projects as well as SkySQL.

https://github.com/mariadb-corporation/mariadb-enterprise-columnstore-docker/blob/master/docker-compose.yml#L30-L34

In this case, CMAPI-6.4.1 refuses to continue unless auto_failover is set to false but even then it puts the cluster into an unusable state. Columnstore will not even complete a simple table creation.

I don't think this local storage check was needed in the first place since we do not recommend customers to use local storage for metadata under any circumstance and we do not sell the product in that way.

I would like these changes backed out and the error message noted in MCOL-5157 removed.

All of this is made moot by MCOL-4218 anyway

 Comments   
Comment by alexey vorovich (Inactive) [ 2022-07-14 ]

yep.. lets wait for dleeyh and toddstoffel to provide an exact repro
then dev will look at it

Comment by alexey vorovich (Inactive) [ 2022-07-14 ]

dleeyh alan.mologorsky do we have a reproduction here and how is it different from the stand docker compose we have been running all along

Comment by alexey vorovich (Inactive) [ 2022-07-15 ]

toddstoffel my initial reaction is that we need to protect ourselves from this .
So when you built the 614 handoff the upstream server latest version moved .
should we adjust process this moving forward ?

Generated at Thu Feb 08 02:55:46 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.