Release 2.2 - Change the queue implementation with Jira
- Adrien Ragot
- Mileva Briand
This release contains changes which are necessary for the performance in cluster environments and for the Data Center certification. Among those, administrators must add some credentials in the Requirement Yogi administration for the integration with Jira. There is no "actual feature" in this release.
❗️ Please set the administrator credentials
When requirements are changed on the Confluence side, we create messages in a queue that is sent to Jira. The queue used to be automatically sent using threads, but this is not regular enough and we've transformed the queue into a job.
Important We need some credentials for this job, so that it can connect to Jira.
Go to the Requirement Yogi administration,
Tab "Integrations",
Enter the username of a user who can modify any "Remote Issue Link" on the Jira issues that are linked to Confluence,
Save.
important For the moment, when the links of an issue in Jira are updated to reflect the changes in the requirement, the change is attributed to the credentials you've provided above. As a reminder, this was necessary because messages from Confluence are processed in batch and allows us to gain performance.
What are those credentials for?
AUTHOR IS LOST: Modifications in Jira will be performed under this username, not under the name of the person who performed the change.
IN A NEAR FUTURE: Next month, if you upgrade the Jira side to a version 2.2.x that we'll publish for Jira, then Jira will display the real author of the changes in the history panel (as displayed in Confluence at the time of editing), instead of this username.
Should I be worried that my login is being used here?
No. It also fixes a bug where changes would sometimes be attributed to a wrong user anyway. At least, in our situation, changes are consistently attributed to one administrative user; and in a near future, Requirement Yogi in Jira will display the Confluence user who performed the changes.
It is limited to only change the RIL ("Remote issue links") in Jira, and only those of type "Requirement Yogi".
You're not giving "us" any information by providing your login here. If we just wanted some login, we could get it from the database.
What happens if I don't provide a login?
The message queue fills up, the table AO_32F7CE_AOINTEGRATION_QUEUE will become large, and there will be messages in the Confluence logs.
Other changes
We've managed the queue (of messages to Jira) as a scheduled job. You can now configure how often this queue executes, you can disable it or run it on demand.
We've changed the caches for the Dictionary blueprint, for the baselines, for some background tasks and their error reporting, so that all of those caches are properly shared on a cluster. We've also measured their performance in preparation for the Data Center certification.
Changes in the RY for Jira module
It is better to also upgrade the Jira module. Using this module, changes will correctly be attributed to the original author. If you remain below version 2.2 in Jira, changes will be attributed to the user provided in the form above.
We've summarized everything about the Jira migration on this page.
RY-107 We've fixed the "bodyfields" box.
List of Jira issues
See the list of Jira issues here.