Was running the trial which expired over the weekend. Sat down today to renew it, then refreshed my WebUI, and all my dashboard statistics are gone.
The payment portal makes you back pay the missing days to when the trial expired, but it dumps all of your statistics.
Why not keep stats for what, a week even? Possibly implement an auto-renew function in the payment portal so the licence doesn’t expire? Business happens.. we had a holiday in the middle of the week and so we’re out busy accommodating customers all week, not clock watching.
When you pay annual, what’s a few days of grace as a consideration?
Cheers
Edit: Moving to issues because now statistics are fully broken.
Coming back to this a day later, and I still have zero statistics, so not only is old data gone, nothing new is being logged either.
I checked the Metrics and they are still enabled with nothing excluded, so this has turned into an unhandled event of some sort, or I am completely clueless on what has happened.
One of my cluster nodes was still operating under the idea it was expired. Now that they all are back up to speed the statistics have reappeared.
If one node pulls an updated licence, shouldn’t that be reflected in the operating state and they’d all know about it?
My problem is resolved - if this is intentional behaviour in a cluster or a bug, I don’t know.
Enterprise metrics recording is gated per node on the license being valid, so a node still holding the lapsed license behaves as expired (stops contributing and serving stats) until it reloads the renewed one. Nothing was actually deleted, which is why the data reappeared.
Is a manual restart always required on all nodes, or will nodes eventually sort themselves out?
A manual restart isn’t strictly required. Each node runs a scheduled license-renewal task that kicks in within about four days of expiry, re-fetches from the license server using the stored auto-renew API key, and on success reloads its enterprise config and broadcasts a reload to the rest of the cluster, so the other nodes converge without being restarted.
The case where a node stays stuck is when its renewal never runs (no auto-renew key configured, or it exhausted its retry window while fully lapsed). There a restart is just the fast way to force a re-validation. Nothing is ever deleted; metrics recording resumes per node as soon as each node holds a valid license again, which is what you saw when the lagging node caught up.