Storage Icon WebUI Update Takes 2-5 Minutes when Storage Node Is Off-line
-
Server
- FOG Version:1.5.0 RC3, today’s working branch on all FOG servers
- OS: CEntOS 7
Description
I found that when a storage node is configured and not working, the webUI Storage Icon takes a long time to update. This delay occurs when you add a new storage node that isn’t connected or if you shut down a storage node that was previously working fine.
I never connected the new storage node since it was just a test, but the problem more or less goes away when you restart all defined storage nodes. I say more or less because the time to update the screen after clicking the Storage Icon can be 15+ seconds when the nodes are up and working with just 2 storage nodes. If the code in the WebUI is serially contacting each storage node, the normal update time could be lengthy if you have a dozen storage nodes defined (which we will).
Jim
-
Sorry - I misspoke -
It seems the problem is with the entire UI, not any particular action. Having a storage node off appears to negatively impact the entire WebUI. We’re seeing delays of 2 minutes, then some brief time of regular responsiveness, then 2 minute delays, across the WebUI.
Specifically, hit storage icon, waited 2 minutes, then had enough regular responsiveness to click hosts, then groups, then Snapins. when I click on a particular snapin, it took another 2 minutes to come up.
Restarting the down storage node returns the WebUI to normal responsiveness.
Jim
-
I noticed this last night and have been working on a fix for the “delay” as well.
if you’d like, you can re-pull the working branch and see if it helps you out.
-
@Jim-Graczyk The points of “unresponsiveness” is not the GUI code directly. It’s because of the call that started the 2 minute load time to occur.
-
We’ll update at lunch time. We just ran the update at about 9am Eastern and we need to get work done on the project. We’re good as long as we don’t shut down storage nodes…
Do want to point out that even with all nodes up, the time after one clicks on a snapin to edit it and the time after clicking on the Storage icon are much longer than other actions and seems unrelated to the amount of data being pulled (heck our database dump was 500K so it’s certainly all in RAM on the server).
Jim
-
Storage Icon should be a little better. Part of this delay is because of using the backend to get the data. The backend pulls in all the files from the nodes (even if they’re not needed). I could tweak the code base, but I think using the api to get the data is ultimately a smarter method as it makes things “common”.
