1000 3925710 0.0 0.0 10552 2312 ? S 19:40 0:00 | | | \_ sh -c git merge-base --is-ancestor e20c10c056d2ed11533c3428e33e93cb31fa22d8 HEAD 2> /dev/null; echo $?
1000 3925711 92.4 15.1 710404 590864 ? R 19:40 0:24 | | | \_ git merge-base --is-ancestor e20c10c056d2ed11533c3428e33e93cb31fa22d8 HEAD
1000 2170558 0.1 10.3 6573012 402832 ? Sl Aug17 35:21 | | \_ pitchfork (gen:0) worker[1] - requests: 75111, processing: /
1000 3925725 0.0 0.0 10552 2316 ? S 19:40 0:00 | | | \_ sh -c git merge-base --is-ancestor e20c10c056d2ed11533c3428e33e93cb31fa22d8 HEAD 2> /dev/null; echo $?
1000 3925726 90.1 14.4 710404 562628 ? R 19:40 0:15 | | | \_ git merge-base --is-ancestor e20c10c056d2ed11533c3428e33e93cb31fa22d8 HEAD
1000 2170585 0.0 8.3 6578772 327676 ? Rl Aug17 10:28 | | \_ pitchfork (gen:0) worker[2] - requests: 8139, processing: /t/two-upcoming-books-about-videogame-history/4406
1000 2170608 0.0 7.8 6589908 305532 ? Sl Aug17 8:08 | | \_ pitchfork (gen:0) worker[3] - requests: 4055, waiting
1000 3925781 0.0 0.0 15992 2684 ? S 19:40 0:00 | \_ sleep 1
```
First seen about 2 hours ago I think.
---
### Author:  [@Ed\_S](https://meta.discourse.org/u/Ed_S)
#### Post date: [September 2, 2026, 7:48pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/5 "2026-09-02T19:48:43Z")
So
`DiscourseUpdates.has_unseen_features`
seems to be a feature which is bringing in a realtime dependency on github working correctly? Even when we’re not trying to upgrade?
---
### Author:  [@Ed\_S](https://meta.discourse.org/u/Ed_S)
#### Post date: [September 2, 2026, 8:30pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/6 "2026-09-02T20:30:46Z")
Edit: this did not solve my problem, as I discover after several hours.
I’ve unchecked this feature:
admin-\>site admin → version checks → clear the checkbox → save
> Site admin
> Configure various settings relating to the management of your site
> version\_checks
> Ping the Discourse Hub for version updates and show new version messages on the admin dashboard
---
### Author:  [@jeanas](https://meta.discourse.org/u/jeanas)
#### Post date: [September 2, 2026, 8:42pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/7 "2026-09-02T20:42:04Z")
How did you manage get the admin interface to open without error in the first place?
---
### Author:  [@Ed\_S](https://meta.discourse.org/u/Ed_S)
#### Post date: [September 2, 2026, 8:47pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/8 "2026-09-02T20:47:12Z")
I had to wait until the git nonsense timed out, and/or could get my request served to a pitchfork which wasn’t blocked.
My forum/logs area is presently showing 4 cases of “pitchfork worker timed out” and it’s not a very busy forum at all.
---
### Author:  [@Ed\_S](https://meta.discourse.org/u/Ed_S)
#### Post date: [September 2, 2026, 8:53pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/9 "2026-09-02T20:53:03Z")
Edit: this did not solve my problem, as I discover after several hours.
One could also use rails from an ssh access, I believe.
This might be a terrible idea. Proceed at your own risk.
```plaintext
# docker exec -i app rails runner 'puts SiteSetting.version_checks'
true
# docker exec -i app rails runner 'SiteSetting.set("version_checks", false)'
# docker exec -i app rails runner 'puts SiteSetting.version_checks'
false
```
---
### Author:  [@satonotdead](https://meta.discourse.org/u/satonotdead)
#### Post date: [September 2, 2026, 9:20pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/10 "2026-09-02T21:20:35Z")
It seems that Github continue with issues, that were the previous ones:
> **[The August 17 outage, and the work ahead](https://github.blog/news-insights/company-news/the-august-17-outage-and-the-work-ahead/)**
>
> An update on the August 17 outage and the steps we're taking to improve reliability.
---
### Author:  [@Ed\_S](https://meta.discourse.org/u/Ed_S)
#### Post date: [September 3, 2026, 7:10am UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/11 "2026-09-03T07:10:29Z")
> [@Ed\_S](#):
>
> My forum/logs area is presently showing 4 cases of “pitchfork worker timed out” and it’s not a very busy forum at all.
I’m now seeing 15 cases of pitchfork worker timed out, with two different backtraces. On the linux side pitchfork continues to accrue unreaped children and sometimes I see a git process which is - I think - the culprit.
My change is not a fix and might or might not be helpful. I will revert it.
> [@Ed\_S](#):
>
> I’ve unchecked this feature:
>
> admin-\>site admin → version checks → clear the checkbox → save
github themselves are [still broken](https://github.com/orgs/community/discussions/206581). The change to HTTP/1.1 for git connections might be a good workaround.
> [@Fatal: could not read Username for 'https://github.com'](https://meta.discourse.org/t/fatal-could-not-read-username-for-https-github-com/411498/5):
>
> It seems to be an issue with Github, found a discussion here
---
### Author:  [@satonotdead](https://meta.discourse.org/u/satonotdead)
#### Post date: [September 3, 2026, 11:17am UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/12 "2026-09-03T11:17:00Z")
Confirmed, they are still having issues in their systems:
`fatal: remote error: GitHub is temporarily limiting some unauthenticated downloads to protect the stability of the platform. Please retry later or authenticate.`
`could not read Username for '``https://github.com/``': No such device or address`
This time not published on [https://www.githubstatus.com/](https://www.githubstatus.com/) but with the same impact as the August incident.
Perhaps it is time to move to Gitlab.
---
### Author:  [@Ed\_S](https://meta.discourse.org/u/Ed_S)
#### Post date: [September 3, 2026, 11:36am UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/13 "2026-09-03T11:36:20Z")
> [@satonotdead](#):
>
> they are still having issues in their systems
On [the github issue](https://github.com/orgs/community/discussions/206581) it seems people (at least in Europe) are seeing the problem is resolved. It seems OK for me (my servers are in Finland)
Anyone confirm they still have a problem?
---
### Author:  [@jeanas](https://meta.discourse.org/u/jeanas)
#### Post date: [September 3, 2026, 12:01pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/14 "2026-09-03T12:01:58Z")
My own forum is working again as well. However, the issue is certainly not solved in general: an outage of GitHub shouldn’t cause an outage of Discourse forums in the first place.
---
### Author:  [@Jagster](https://meta.discourse.org/u/Jagster)
#### Post date: [September 3, 2026, 12:15pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/15 "2026-09-03T12:15:08Z")
Yet it does, because any reason why rebuilding of a container fails, will kill Discourse (and that’s why we have an option: 2-container setup).
But sure. It would be nice if the app-container would survive.
---
### Author:  [@Ed\_S](https://meta.discourse.org/u/Ed_S)
#### Post date: [September 3, 2026, 4:05pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/16 "2026-09-03T16:05:52Z")
> [@jeanas](#):
>
> an outage of GitHub shouldn’t cause an outage of Discourse forums in the first place.
I completely agree. And in my case, there was no rebuild or other intentional git action. These were I think ‘helpful’ git actions which turned out not to be helpful.
I would like my self-hosted forum to run unattended even if github (and [discourse.org](http://discourse.org)) fell into a very deep hole.
---
### Author:  [@Falco](https://meta.discourse.org/u/Falco)
#### Post date: [September 3, 2026, 4:52pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/17 "2026-09-03T16:52:06Z")
> [@Ed\_S](#):
>
> I would like my self-hosted forum to run unattended even if github (and [discourse.org](http://discourse.org)) fell into a very deep hole.
💯
It is completely unacceptable to be any other way.
---
### Author:  [@Jagster](https://meta.discourse.org/u/Jagster)
#### Post date: [September 3, 2026, 4:54pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/19 "2026-09-03T16:54:28Z")
> [@Ed\_S](#):
>
> I would like my self-hosted forum to run unattended even if
And it will — unless an admin says `.launcher rebuild app`. Or what am I missing now?
---
### Author:  [@Falco](https://meta.discourse.org/u/Falco)
#### Post date: [September 3, 2026, 4:55pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/20 "2026-09-03T16:55:48Z")
@Ed_S I think I found the issue and I have a PR for it, can you check to see if it touches the area you saw with issues?
>
>
> Previously, \`CurrentUserSerializer#has\_unseen\_features\` could derive the newest …new-feature timestamp inline, which filters the feed through \`GitUtils.has\_commit?\` — and since we deploy from a partial clone (\`--filter=tree:0\`), \`git merge-base --is-ancestor\` on a commit missing from the local graph is lazily fetched from the promisor remote, i.e. a blocking, untimed network call to GitHub, once per feed entry, in a request. Compounding it, a fully-filtered feed cached no result at all, so every logged-in page render redid the whole loop until the worker timed out.
>
> This change runs that check with \`GIT\_NO\_LAZY\_FETCH=1\` under a hard timeout so it stays local and bounded, memoizes it per process, and makes \`DiscourseUpdates.latest\_new\_feature\_created\_at\` a plain Redis read derived on write (\`update\_new\_features\`, plus a throttled \`Jobs::RefreshLatestNewFeature\` to self-heal after a deploy or an upcoming change changing status).
---
### Author:  [@Ed\_S](https://meta.discourse.org/u/Ed_S)
#### Post date: [September 3, 2026, 4:58pm UTC](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513/21 "2026-09-03T16:58:55Z")
> [@Falco](#):
>
> [FIX: Stop new-feature checks reaching GitHub during requests - Pull Request #43143 - discourse/discourse - GitHub](https://github.com/discourse/discourse/pull/43143)
oh, nice, that does seem to me to have the right shape! (Edit: I read the commentary, not the code!)
For interest, when you say per process, is that per pitchfork?
[Next page](https://meta.discourse.org/t/error-for-logged-in-users-and-git-hang/411513.md?page=2)