Why does it takes so much time to create a comment after updating Jira and Jira Service Desk

[Hannes]

After the update of Jira and Jira Service Desk we have the problem that in some Jira projects it takes a very long time until a comment is created or a process is closed/opened.

Not all projects are affected. And in the projects that are affected, the problem is different.
For example, in one project the creating or closing is very slow and the rest is ok. But in another project the creating is normally fast, but changing the status or commenting takes an extremely long time.

We run Jira and Jira Service Desk on a Ubuntu 14 LTS.

We Update Jira from 7.6 to 7.12.1 and Service Desk to 3.15.1

Dateabase: MYSQL 14.14

What other information do you need?

[KellyW · Accepted solution]

Hi Hannes,

If the performance issue is pertain to specific project, I'm leaning towards custom fields and workflow specific to this project. If there's any problem with these, it should be logged in atlassian-jira.log.

Do test in Safe Mode to rule out any third-party add-ons.

Then, identify the transition that triggered the slowness and check on the settings in the workflow.

Regards,
Kelly

[Hannes]

Ok, I'll try it later today. Thank you for your reply.

Regards
Hannes

[Jason Michaud]

Did you ever find an answer to this. We experience similar issues when creating new issues (10-15 secs) and transitioning issues (5-8 secs) from one status to another. We are currently on Service Desk 3.10.2 and Jira Core 7.7.2

[Hannes]

Hello Jason,
Sorry for the late reply. We were able to locate the problem. We use the plugin "Mobile for Jira". The plugin itself does not cause any problems. But as soon as you have selected a scheme in a Jira project under "Mobile for Jira Notification", all actions that are carried out within the project (change status, write comments,...) are delayed.
We changed the schema to "select" and with that all mentioned problems were solved immediately.

[Jason Michaud]

Hi Hannes

Thank you for your reply. That is interesting. We use Mobile for Jira as well and use our Service Desk notification scheme for our Mobile for Jira notification scheme.

Early in May - we finally figured out how to get around all the performance issues we'd been having for 10 months. One of our DBA's was testing around in our test environment - he added a new index to the entity property table. Once he did that - all out performance issues went away.

But since I just saw your most recent reply - I went ahead and updated the notification scheme to be Select for now to see if it benefits us in any way. Let me know if you have any questions on what we did.

Thank you!

Jason

[Hannes]

Hi Jason,

Thanks for your hint with the entity table. Can you explain what he have done? Only rebuild the table and the indexes?

What kind of server and db you use? Our Jira runs on a vm with linux and mysql.

Kind regards,

Hannes

[Jason Michaud]

Hi Hannes

We created a net-new index that covered our query more appropriately. We did not rebuild the table nor the index, we added a new index.

The index we created has the following columns as the index key: "ENTITY_NAME", "ENTITY_ID", "PROPERTY_KEY". It closely matches one of the ones Atlassian recommends but is slightly different.

This solution seems to have worked for our environment - i'm not sure you will have the same result. My DBA says you should monitor the Active Expensive Queries (or Recent Expensive Queries) and see which query is causing the issues. From there you can look at the execution plan and try to come up with an index or resolve it best they can. We are on SQL Server 2014 (SP3) Standard Edition

Hope this helps,

Best

Jason