v1.3.4 UI Regression (possible state management regression)

[djenriquez]

Nomad version

nomad version
Nomad v1.3.4 (a267b2b000ae72ba05fa571eaafe170a7b9c4d1b)

Operating system and Environment details

Amazon Linux 2

Issue

After upgrading from 1.3.3 to 1.3.4, seeing some regression in the UI:

1. The server page is not properly identifying the leader (every node has leader:false even though the /v1/status/leader endpoint reveals the leader), nor is it clearing out the old 1.3.3 servers that have left even after 24 hours (the data from v1/agent/members does not seem to be clearing out the old servers)

2. In our development environment where ACLs are disabled, the "exec" button is disabled, even though you can click this button and an exec pop up comes up.
   

3. Related to 2, after the exec pop up window comes up, you are unable to actually run the command to execute the exec. The window just hangs. Chrome dev tools show no network traffic after pressing enter, its as if its just disabled completely.
   

4. Also seeing a js error in the console (hard to tell if these are expected since JS errors are so common these days):

Failed to load resource: the server responded with a status of 400 ()

nomad-ui-2337cfef252f8cd23727e6c21be6e3d8.js:2076 Uncaught TypeError: Cannot read properties of null (reading 'includes')
    at get activeRegion (nomad-ui-2337cfef252f8cd23727e6c21be6e3d8.js:2076:10)
    at vendor-eb07b60f7e1f309e783749d1ba18f9d7.js:2000:67
    at e.untrack (vendor-eb07b60f7e1f309e783749d1ba18f9d7.js:4318:12)
    at Re.get (vendor-eb07b60f7e1f309e783749d1ba18f9d7.js:2000:56)
    at z.r [as activeRegion] (vendor-eb07b60f7e1f309e783749d1ba18f9d7.js:1964:245)
    at r.getTaskStateSocket (nomad-ui-2337cfef252f8cd23727e6c21be6e3d8.js:2051:179)
    at h.openAndConnectSocket (nomad-ui-2337cfef252f8cd23727e6c21be6e3d8.js:1141:844)
    at e.default.handleDataEvent (nomad-ui-2337cfef252f8cd23727e6c21be6e3d8.js:2493:60)
    at nomad-ui-2337cfef252f8cd23727e6c21be6e3d8.js:2492:130
    at e.fire (chunk.331.2bf16a8aab827bcd7550.js:786:29)
get activeRegion @ nomad-ui-2337cfef252f8cd23727e6c21be6e3d8.js:2076
(anonymous) @ vendor-eb07b60f7e1f309e783749d1ba18f9d7.js:2000
e.untrack @ vendor-eb07b60f7e1f309e783749d1ba18f9d7.js:4318
get @ vendor-eb07b60f7e1f309e783749d1ba18f9d7.js:2000
r @ vendor-eb07b60f7e1f309e783749d1ba18f9d7.js:1964
getTaskStateSocket @ nomad-ui-2337cfef252f8cd23727e6c21be6e3d8.js:2051
openAndConnectSocket @ nomad-ui-2337cfef252f8cd23727e6c21be6e3d8.js:1141
handleDataEvent @ nomad-ui-2337cfef252f8cd23727e6c21be6e3d8.js:2493
(anonymous) @ nomad-ui-2337cfef252f8cd23727e6c21be6e3d8.js:2492
e.fire @ chunk.331.2bf16a8aab827bcd7550.js:786
(anonymous) @ chunk.331.2bf16a8aab827bcd7550.js:787
e.fire @ chunk.331.2bf16a8aab827bcd7550.js:786
t.triggerDataEvent @ chunk.331.2bf16a8aab827bcd7550.js:1502
t._keyDown @ chunk.331.2bf16a8aab827bcd7550.js:221
(anonymous) @ chunk.331.2bf16a8aab827bcd7550.js:181

<REDACTED>:1          Failed to load resource: the server responded with a status of 500 ()

allocations:1          Failed to load resource: the server responded with a status of 500 ()

5. When loading up the jobs page:
   

6. The UI no longer displays the Nomad Version on the bottom left hand corner (bottom of the navigation menu).

7. Clients that have rotated out are no longer being GC'd, atleast not as the same rate as before, hundreds of clients report as down (we have a busy cluster). This screenshot is filtering "Down" status.
   

Reproduction steps

Upgrade Nomad from 1.3.3 to 1.3.4.

[djenriquez]

I found other similar issues after posting, I do have some cases not discussed in those issues though. Should I keep this issue open for that or just close and post the additional info over on the existing issues?

[aniaptebsft]

Seeing the same issue with grayed out exec button. This is causing a lot of disruption to our engineering workflows.

[jrasell]

Hi @djenriquez and @aniaptebsft and thanks for raising this with the context. I will link the issue to #14359 which has been tracking some UI regressions within 1.3.4, however, I will let @philrenaud read through this and decided whether to close this issue out as duplicate.

[shantanugadgil]

Hi, we are seeing a bunch of the UI issues reported after upgrading to 1.3.4 for the Nomad servers.

This particular issue (Exec button disabled) is especially bothersome.

The UI shows the version a the bottom left only on the first load, and then it disappears.

cross referencing #14359 #14331

[shantanugadgil]

Another question @jrasell @philrenaud how concerned should we be with version 1.3.4 and the non-UI functionality. Have there been any issues reported which are not related to the GUI?

The current state is that we have upgraded ONLY the servers to 1.3.4 and were planning to roll out the clients with 1.3.4 within the week.

[philrenaud]

Have there been any issues reported which are not related to the GUI?

Hi @shantanugadgil — the issues described are solved in #14381 which are restricted to the UI alone. We would still recommend moving to 1.3.5 when that releases (very shortly)

[philrenaud]

Hi @djenriquez , @aniaptebsft — sorry for the hassle here. The issues you see will be fixed with 1.3.5, going out shortly, and can temporarily be alleviated on a per-browser basis by adding the following to your browser console:

localStorage.setItem('nomadTokenSecret', 'thisisathirtysixcharacterlongstring!'); (per this comment)

Closing this issue and tracking in #14359

[shantanugadgil]

This is good news indeed. Thanks! 👍🏻🙏🏻

[github-actions]

I'm going to lock this issue because it has been closed for 120 days ⏳. This helps our maintainers find and focus on the active issues.
If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.