We run two separate instances of Pulse, one on our prod server, one on our dev server.
Yes, yes, I know that you can run everything on one instance. No, I don’t want to do that. No, the reason is not germane. (But the fact that it allows me to do a test upgrade on the dev server first shows one good reason.)
We were running version 7.3 on Dev (and still are on Prod). This morning I upgraded Dev to 8.0.0.
Everything seemed OK, aside from two things:
(a) An error dialog during the upgrade process (which took care of uninstalling 7.3) which read:
Error opening file for writing:
C:\Program Files\Pulse for TM1\bin\pulse-dashboards-service.exe
Options
Click Abort to stop the installation.
Retry to try again, or
Ignore to skip this file.
That one I’m not so worried about because I’ve seen it in previous upgrades and ignoring it made no difference that I could detect.
The new one is that under Administration => Alerts… every single alert was obliterated. Memory, Server Status, Disk Space… every one is gone. The User accounts are still there, the notification groups are still there, but there is not a single alert left.
I recreated one of the “server offline” alerts and dropped the server; it worked, so there is nothing wrong with the alert system itself… it’s just that I have to recreate all of them.
Has anyone else had the same experience?
Hi @AlanKirk ,
Great to see you keeping up with the latest releases. And yes, I agree with you: even though a single Pulse instance could be enough, having two is safer, as it lets you test an upgrade before touching your Pulse production server.
When upgrading to Pulse v8.0.0, it looks like a simple upgrade since you don’t even need to upgrade the Pulse Monitor, but in the back end the H2 database is being upgraded too. It’s a major upgrade, so a data migration takes place.
The database contains many tables. Since you’re saying that “only” the alerts are missing and that your Pulse users are still there, it looks like something went wrong when the Alerts table was migrated.
The migration has completed and Pulse 8 is working now, so if you’re happy to continue and recreate the alerts manually (or use the Pulse Alert Templates), you can simply keep it as is on your dev server.
That said, it would be good to understand what happened so it doesn’t repeat itself on your production server. Could you please raise a ticket and share your Pulse logs here: Jira Service Management
Thanks for sharing this on the forum, it’s a good reminder for everyone: always take a backup before a Pulse upgrade. In your case “just” losing the alerts is probably fine, but had the issue been bigger, a backup would have let you roll back to the previous version and restore your data.
Cheers,
Vincent