Created attachment 944024 [details] Maintenance mode and status differences in various application areas Description of problem: Storage Domain Entering Maintenance Mode displays a different status on the main Storage tab, Data Center sub-tab within Storage and Tasks Version-Release number of selected component (if applicable): 3.5 vt4 How reproducible: 40% Steps to Reproduce: 1. Click on the Storage tab 2. Select a Storage domain 3. Click on the Data Center sub-tab 4. Click on the Maintenance button 5. Click on OK to enter maintenance mode 6. Note the status of the Storage domain within the 2 tabs as well as under the Tasks tab Actual results: In some of the cases, it appears that the various tabs are out of sync with each other Expected results: The Storage status should be consistent across the various areas of the application. Note that in some cases, the status was off for a good minute or two
iirc, tasks tab has a separate refresh interval (which is fixed to 5 seconds), hence, there might be a synchronization discrepancy with the other tabs. Regarding main-tab/sub-tab issue, is it specific for the mentioned set of tabs (e.g. is it reproduced in 'Data-Centers main-tab -> Storage sub-tab' as well)? @Einav - do we still have any open issues about tabs refresh in infrastructure level?
Created attachment 946773 [details] Data Center tab with Storage subtab inconsistent display when Storage domain goes into Maintenance mode
Please note that I'm seeing this issue with VT5 as well, this is clearly visible under the Storage -> Data Center tab (where the top area can be out of sync for 10 or more seconds). I've checked the main Data Centers -> Storage tab and the situation there is reversed. When setting Maintenance mode on a domain, the Data Centers portion immediately shows the Storage domain is in Maintenance while both the Storage tab and Tasks lag behind upto 10 seconds in my tests. Please find the attached screenshot
(In reply to Daniel Erez from comment #1) > iirc, tasks tab has a separate refresh interval (which is fixed to 5 > seconds), hence, there might be a synchronization discrepancy with the other > tabs. Regarding main-tab/sub-tab issue, is it specific for the mentioned set > of tabs (e.g. is it reproduced in 'Data-Centers main-tab -> Storage sub-tab' > as well)? > > @Einav - do we still have any open issues about tabs refresh in > infrastructure level? all tab-refresh infrastructure issues were already resolved. all data-sync-"lags" should be extremely short, even unnoticeable, in most cases. from comparing attachment 944024 [details] and attachment 946773 [details], it seems that no matter where you are in the GUI, the Data-Center data seems to be up-to-date ("Maintenance") while the Storage-Domain data is "lagging" ("Preparing for Maintenance"). therefore, I suspect that the data itself (i.e. on the engine side) is out-of-sync, and the GUI merely reflects this data. I would test the REST API to verify that (i.e. perform the action, query the rest-api for the Data-Center, query the rest-api for the Storage, etc.), and would investigate on the engine/db-update side - maybe something is problematic there.
I'm guessing the problem is fn_get_storage_domain_shared_status_by_domain_id which is implemented in a simply wrong fashion, which is both slow and racy. Since this bug is hard to reproduce and the solution will not be easy, pushing out to 3.6.0 to reevaluate once we've removed the pool.
Target release should be placed once a package build is known to fix a issue. Since this bug is not modified, the target version has been reset. Please use target milestone to plan a fix for a oVirt release.
Benny, this issue is most likely fixed, please try to reproduce
I couldn't reproduce this, last build commit: 550a653f7cbc72ac8f414113e541a3e89ee0a6c5
This was probably fixed somewhere along the road, moving to ON_QA so the QE guys can confirm
Verified on rhevm-4.1.1-0.1.el7