PMM2 Dashborad session timeout

If I set the “Home Dashboard” to any time range other than the default, after about 5 min the browser session times out. And if I set the Dashboard to anything other than the “Home Dashboard” (eg OS>Network Details) it again times out after 5 min. The only way I, so far, I’ve found to keep it from timing out is to go to the default timeframe for the “Home Dashboard”.
I looked on Grafana’s forum and it appears there was a similar issue/bug early in the 6.x versions like 6.2.ref:
I’d like to know if there’s a way to keep this from timing out as it’s a bit of an inconvenience to login every 5 min.
PMM ver: 2.9.0

Actually I was able to get the default “Home Dashboard” to timeout after 10 min. Which is better than 5 min for all the other dashboards, but still no bueno.

Hmmm…shocked to see this from 2.9.0…I’ll give this a try over the weekend…admittedly I rarely change the time range on the default dashboard and for the pages I do change I don’t stay there long.  I assume when you say “timeout” you mean are logged out and have to relogin to the UI or are you seeing something else like an error?  

Exactly, it times out in the UI and users have to log back in again.

I cannot for the life of me reproduce this…I have set the time window from 12hrs to other defaults (2 days, last 15 minutes, etc)  and to several custom ranges (now-37 min, two random days) and cannot get it to timeout.  I’ve set the page refresh to all kinds of different values as well but I stay logged in at all times.  Is there any “cookie killer” or something running on your machine?  Maybe an incognito window?  I’ve encountered some companies that set some policies around chrome and when users logged into chrome as their company gmail account unexpected things were happening…  I have no other ideas but if you think this is on our end, please submit a bug against PMM to and we’ll get all the vital data so it can be more properly investigated. 

No add-ons or extensions added to Firefox browser which is running on an ubuntu 20.04LTS desktop. I’ll spin up a Chromium browser and test with that. I’ve also tested with the latest Firefox on my Apple macOS 10.15.5 laptop.
It’s starting to point to my docker container for PMM.
I’ll report back what I find

I tested in Chromium with the same result. One difference probably between your test and mine is you’re probably using a local account created within PMM, and I’m using LDAP auth integration in Grafana. I started to think it has something to do with PMM image and specifically Grafana+LDAP auth. So then I tested with a local account and saw the same issue. Taking a look from Grafana admin side it looks like the session is still active, but from the client it’s at a login page for Grafana (https://fqdn/graph/login). And each time I login it creates a new session(from within Grafana’s Users Session page) and the old sessions seem to accumulate without ever timing out to be killed or forced logout after X amount of inactivity time. I’m not too worried now about 5min of inactivity and then the users have to log back in, but more concerning is the sessions count and potential resource exhaustion where the PMM container needs to be restarted to clear all the old sessions.

I WAS ABLE TO REPRODUCE THIS!!! Don’t ask me why I am so excited…but wanted to validate you are NOT crazy ;-)!  I’m also on a 2.9.0 system (will narrow down the steps and see if it’s still happening on the newly released 2.9.1) running on CentOs 7.  On the system I was testing with it is a vanilla install (although integrated with AD for auth and with SMTP enabled for alerting).  I made the “Home Dashboard” different than default.  I log in to PMM, immediately changes screens to the original home dashboard and let it sit for about 15 or so minutes and when I try to do something i saw 3 red boxes flash up quickly and was logged out.  I’ll play a little more with this as time permits but should be able to put enough information together to submit a bug.  If you PM me your email I’ll make you the reporter or watcher so you can get alerts on it’s progress.