When Jira runs out of database connections, it has no option but to stop responding - it can't actually work without the data from the database.
It is worrying that you are using up your pool, it suggests something is going wrong and not releasing connections back into the pool. It may also be poor configuration, but if you've installed Jira off the shelf, and you've not taken the size of it up to thousands of users with huge numbers of issues, that is unlikely.
There are three things you can look at:
1. A very blunt "increase the number of connections in the pool".
If you're simply running out because of load, then this could easily fix the problem (although if it does, I'd still recommend tuning properly).
This is worth a try as a debugging run - let's say you double the connections in the pool, does the unresponsiveness stop? Or does it drop to a few seconds before it re-awakens? Does it still happen, but after 6-8 hours instead of 3-4? Or does it not really change? Each one of those can tell you something more about the problem.
2. Tune the pool
Probably should be done on any medium-sized or larger production system no matter whether you've got a problem with it or not. https://confluence.atlassian.com/adminjiraserver/tuning-database-connections-938846864.html
3. Read the logs - if there are sudden spikes in connection usage, it's likely that the logs will have warnings or errors in them that could tell you what is chewing up all the connections suddenly. They probably won't tell you the exact problem, but they are likely to point you at a specific process, a specific function in Jira, a corrupted database record, or a misbehaving app.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.