[MCOL-4227] Multinode - doesnt work after the stop and start commands were executed Created: 2020-08-03  Updated: 2021-05-08  Resolved: 2020-09-11

Status: Closed
Project: MariaDB ColumnStore
Component/s: cmapi
Affects Version/s: 1.0.0
Fix Version/s: 5.4.1

Type: Bug Priority: Major
Reporter: David Hill (Inactive) Assignee: Todd Stoffel (Inactive)
Resolution: Fixed Votes: 0
Labels: None
Environment:

Multi-node system using the Ansible procedure

Attachments: PNG File Screenshot from 2020-09-10 12-58-48.png    
Sprint: 2020-8

 Description   

Reported by customer:

We set up a multi-node system with the ansible instructions that were on the git using our own servers. We verified the system was running for Columnstore with multiple nodes, using storage manager and the S3 for storage. We tried the stop and start of the cluster commands as shown in the git (using curl commands). When these came back up, the system was in a read only state and not stable. Any columnstore table that tried to be queried would point to it not existing in the Columnstore engine, which indicates an issue with the dbrm or frm files. I checked that the frm exists for this. It is possible that further documentation that will be released may address some of these issues.

They reported it looks similar to MCOL-4226, but 4226 couldnt get the system going and in this issue they achieved a good install and startup, but it failed to work after stop/start command

 Comments   
Comment by Jose Rojas (Inactive) [ 2020-08-27 ]

This is fixed in new version of CMAPI, which will be available upon next enterprise release.

Comment by Jose Rojas (Inactive) [ 2020-08-27 ]

Will be available to test once new version of cmapi is available, at which point ansible playbook will be updated as well to test in this exact environment

Comment by Jose Rojas (Inactive) [ 2020-09-03 ]

dleeyh This is unrelated to ansible playbook, and since it has not been updated yet, the proper way to test this would be to setup a cluster with storage manager / s3 and verify that it is functioning after cluster/shutdown and cluster/start commands (can get them from ansible repo).

Comment by Daniel Lee (Inactive) [ 2020-09-10 ]

Build tested: 1.5.4-1 (Drone #598) cmapi (Drone #253)

Created a 3-node cluster using AWS S3
Verified data is being stored on AWS (dbroot directory started empty)
Executed sanity test (DDL, DML, query)
shutdown and start the stack few times
created, loaded, and verified 1gb DBT3 database
Created a 10gb DBT3 database and started cpimport. Aborted cpimport while it was loading the lineitem table. When trying to drop the 10gb database, table lock was in place and operation failed (as expected). Use viewtablelock and cleartablelock to clear the table lock. Drop database was successful.

Comment by Todd Stoffel (Inactive) [ 2020-09-11 ]

This issue is resolved in later builds of cmapi. It will be released in conjunction with 1.5.4

Generated at Thu Feb 08 02:48:46 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.