I just upgraded to 1.5.11, and am encountering an issue. I will say first off I have yet to drop back to a previous version to test, so not sure if this is just this version. I have about 7 different dashboards for a customer now, and they are all working well. I was asked to add to one of them, and noticed that when I restarted the dashboard, ALL dashboards refreshed. I have been testing, and confirmed that this is happening only with a Restart; if I simply Stop one of the dashboards, it does not affect any of the others.
Hi, @adam It is a mixture; 4 of the 7 are set up that way.
Weird, I started a non-AutoStart one, and tested, and it did not refresh when I restarted the other. I also then set that to AutoStart and tested again, and it did not refresh. Seems to be somewhat random.
Having similar behavior. Whenever I edit the PowerShell code for a particular dashboard in the UI and click save, it reloads all dashboards if anyone happens to have one open.
I tried to tune off the AutoStart feature to see if that helped but when I go back in to Edit Properties, the Auto Start is still checked off.
That’s weird. With auto-start off, it shouldn’t be doing that. I’ve opened an issue to investigate this on our end.
One of the reasons this is happening is that with dashboards, if any file changes within the repository folder, the dashboards will restart if you have auto-start on. We do this (maybe incorrectly) because it’s really common for people to have multiple files that define their dashboards. This is different than files like endpoints.ps1 or settings.ps1 that will reload only their properties when those specific files change.
Personally, I think that I would likely suggest not using auto-start for production dashboards because of this issue. The one caveat would be git managed PSU instances that would only be updated when a git sync occurs with changes. I think what we should do within the product is make it clear that a dashboard has been updated but has not been restarted since the last update so you know that what’s deployed to the users does not match what is shown in the code editor. You can then choose to deploy the updated dashboard during maintenance periods etc.
I’m super open to suggestions here but having dashboards randomly restart like this isn’t good for the end user since it loses state and causes weird disruptions that you may not intend to happen.
Thanks @adam, for me that is something I can work with. While I have a number of dashboards that the customer insists on setting to auto-start, for fear of it not being available, I think it provides me with the excuse to go back and do what I should have done in the first place by spinning up a test PSU server for working on new dashboards. That way, if we are hammering a dashboard (like I am currently on a couple) we are not affecting those that are actually being used by end users, and when ready it can be promoted to Production.
@mmcelreath I remember a similar issue, with auto start being set to On even if you chose to leave it off when creating the dashboard, but I thought that was fixed in 1.5.7 or 1.5.8. If I remember correctly, I had to open the dashboards.ps1 file and set the -DisableAutoStart parameter to get it to ‘stick’.
Something we maybe should make a distinction within the the product is auto-start vs auto-restart. Then the dashboards would always be available if the server were restarted or whatever but if you make changes they wouldn’t necessarily restart.
That would be great too. I think that is the concern with this customer; they do weekly patching of their Windows boxes, and a number of apps/services don’t start as they should afterward. And as they are now using one of my dashboards to track those failures, they want to ensure it is up
The service is set to start automatically of course, but if we could make that distinction on a per-dashboard basis that would be awesome.
This would bee great! I have the issue with restarting also. And even if this is a POC it’s not so fun for the users.
As you know during POC we are updating the code many many times during the day.
Will not be a issue during production as then we will have our service windows.
I would highly suggest, even for a POC, that you follow the DEV/PROD methodology. Do all your code writing, with the frequent restarts associated, on a DEV box. Then when you’re confident in a release, copy the code to prod. We were doing it all on one server when I opened this thread; switching to a dev box really eased users’ frustrations, even if they had to wait a few hours or a day for new code to be released.
