Canapin
April 9, 2026, 9:47am
1
It just does this sometimes.
The backup succeeded in this case.
I’m not sure it can also be stuck in this state whenthe backup fails.
Regardless, it always makes me a little anxious when it happens
Nothing in /logs and I didn’t have the dev tools open to check for errors.
edit: if we change the tab and go back to the Logs tab, then they appear normally.
4 Likes
ToddZ
April 9, 2026, 12:24pm
2
I keep forgetting to mention it, but I’ve had this happen too, with each of 4 updates since mid-February.
3 Likes
Ethsim2
July 21, 2026, 6:53am
3
I may have found a possible cause for this.
The backup log MessageBus handler was calling the deprecated Ember Array.clear() method when it received the [STARTED] message. I encountered an error on that path while testing backups and have opened a draft PR replacing it with splice(0):
main ← Ethsim12:fix-deprecated-array-clear
opened 10:39AM - 20 Jul 26 UTC
## What does this change?
Replaces the deprecated Ember native-array-extension … call used when a backup
operation starts:
```javascript
logs.clear();
```
with:
```javascript
logs.splice(0);
```
This clears the backup logs while preserving the existing
`@autoTrackedArray` instance.
## Why?
Starting a backup sends a `[STARTED]` message through the
`/admin/backups/logs` MessageBus channel. Its callback currently calls
`.clear()`, triggering:
```text
discourse.native-array-extensions.clear
```
When JavaScript deprecations are configured to raise errors, this also causes
the MessageBus callback to fail.
## Testing
Added a unit test confirming that a `[STARTED]` message:
- clears existing backup-log entries;
- preserves the tracked-array instance;
- marks the backup operation as running;
- hides the read-only alert.
Local checks completed:
- Prettier
- ESLint
- Git whitespace checks
The focused QUnit test was not run locally because a complete Ruby/Discourse
test environment was not installed.
This appears potentially related to the live Logs tab remaining blank while the persisted logs appear after navigating away and back, but I cannot confirm that it is the same root cause because no browser-console error was captured when this topic was reported.
2 Likes
