Frequent koji server errors when watching builds #13454

New issue

Closed

opened 2026-07-09 16:38:30 +00:00 by catanzaro · 2 comments

catanzaro commented

2026-07-09 16:38:30 +00:00

Description of request

For the past year or so, koji commands have been failing due to unknown server bugs. This never happened until sometime last year, at which point it became very frequent. I'm sorry for not reporting it until now.

For example, today I started a scratch build:

$ fedpkg scratch-build --srpm webkitgtk-2.52.5-1.fc44.src.rpm

After a while, it failed with:

Could not execute scratch_build: HTTPSConnectionPool(host='koji.fedoraproject.org', port=443): Read timed out. (read timeout=43200)

I then decided to manually watch it:

$ koji watch-task 147589090

This, too, eventually failed:

2026-07-09 11:35:04,000 [ERROR] koji: ConnectionError: ('Connection aborted.', ConnectionResetError(104, 'Connection reset by peer'))

Workaround is to open the koji web UI in a web browser and manually refresh the page every once in a while, which shouldn't be necessary.

### Description of request For the past year or so, `koji` commands have been failing due to unknown server bugs. This never happened until sometime last year, at which point it became very frequent. I'm sorry for not reporting it until now. For example, today I started a scratch build: ``` $ fedpkg scratch-build --srpm webkitgtk-2.52.5-1.fc44.src.rpm ``` After a while, it failed with: ``` Could not execute scratch_build: HTTPSConnectionPool(host='koji.fedoraproject.org', port=443): Read timed out. (read timeout=43200) ``` I then decided to manually watch it: ``` $ koji watch-task 147589090 ``` This, too, eventually failed: ``` 2026-07-09 11:35:04,000 [ERROR] koji: ConnectionError: ('Connection aborted.', ConnectionResetError(104, 'Connection reset by peer')) ``` Workaround is to open the koji web UI in a web browser and manually refresh the page every once in a while, which shouldn't be necessary.

kevin commented

2026-07-09 18:00:18 +00:00

Owner

There was another round of scrapers this morning (hitting the wiki this time). They didn't cause any direct outage, but some of the proxies hit max busy workers and when that happens sometimes requests hit timeouts or resets. ;(

Can you see if it's still happening? if not, that was likely it... if so, it's something else.