About using transaction isolation level for my app?

Phong Phạm Gia September 27, 2016

I'm using Galera Cluster in multi-master mode, however it can only use the REPEATABLE-READ level. So can i use REPEATABLE-READ level for JIRA and Confluence application???

1 answer

0 votes
Nic Brough -Adaptavist-
Community Leader
Community Leader
Community Leaders are connectors, ambassadors, and mentors. On the online community, they serve as thought leaders, product experts, and moderators.
September 27, 2016

You cannot cluster JIRA unless you are using JIRA Data Centre (which is a whole different discussion), so this probably isn't going to work the way you think it might.

For MySQL, the applications require a row-level lock (see https://confluence.atlassian.com/adminjiraserver071/connecting-jira-applications-to-mysql-802592179.html ) which generally means you need READ-COMMITTED not repeatable-read.  I don't know enough about repeatable-read to know if it can be used as a row-level lock.

Phong Phạm Gia September 27, 2016

Thanks for your reply & sorry for my bad English, but i mean when i configure Galera Cluster in multi-master mode(MySQL DB) to use for my applications(Jira, Confluence) i must use transaction isolation level is REPEATABLE-READ, cannot use other level. (see http://galeracluster.com/documentation-webpages/isolationlevels.html )

Are there any problems with this configuration? and Can i use other transaction isolation level as REPEATABLE-READ instead of READ-COMMITTED for my applications? 

Nic Brough -Adaptavist-
Community Leader
Community Leader
Community Leaders are connectors, ambassadors, and mentors. On the online community, they serve as thought leaders, product experts, and moderators.
September 28, 2016

Your English is good, but it's a complex language sometimes.  Even as it is my first language, I find it is very easy to misunderstand what people have written.  I think you were very clear.

So, yes, there are problems with your configuration.  The Atlassian applications need to be run with READ-COMMITTED (unless you are using a Data Centre version).  From the documentation you've provided, it's quite clear that Galera does not support that when it is running in multi-master node and hence you cannot use it for your Atlassian databases.  That does make sense to me - JIRA cannot be clustered, and can only really be used with master-slave database clusters, where the slaves are for backups or (ugh) reporting.  (Again, unless you are using JDC)

Suggest an answer

Log in or Sign up to answer
TAGS
AUG Leaders

Atlassian Community Events