go-gitea / gitea

Git with a cup of tea! Painless self-hosted all-in-one software development service, including Git hosting, code review, team collaboration, package registry and CI/CD
https://gitea.com
MIT License
43.11k stars 5.32k forks source link

Timeout when pushing big docker images with docker #31558

Open CPlusPlus17 opened 4 days ago

CPlusPlus17 commented 4 days ago

Description

When pushing a big docker images the command is aborted after 60s.

2024/07/04 15:39:30 ...eb/routing/logger.go:68:func1() [W] router: slow      PATCH /v2/xxx/dotnet/build/blobs/uploads/gfjt0i63h5jxsjle8jgvzbyk5 for 10.244.2.206:57674, elapsed 3517.0ms @ packages/api.go:700(packages.ContainerRoutes.func2.2)
2024/07/04 15:40:27 ...ntainer/container.go:90:apiError() [E] unexpected EOF
2024/07/04 15:40:27 ...eb/routing/logger.go:102:func1() [I] router: completed PATCH /v2/xxx/dotnet/build/blobs/uploads/gfjt0i63h5jxsjle8jgvzbyk5 for 10.244.2.206:57674, 500 Internal Server Error in 60197.8ms @ packages/api.go:700(packages.ContainerRoutes.func2.2)

Traefik log:

10.244.0.0 - - [04/Jul/2024:15:27:53 +0000] "PATCH /v2/xxx/dotnet/build/blobs/uploads/heuof3pvgcekqjxiftepaqqdk HTTP/1.1" 499 21 "-" "-" 223 "websecure-gitea-prd-gitea-ingress-services-xxx-xxx-admin-ch-v2@kubernetes" "http://10.244.4.21:3000" 60000ms

Traefik has a timeout of 600s before aborting the request.

Gitea config:

gitea:
  config:
    server:
      ROOT_URL: https://xxx/gitea-prd/
      PER_WRITE_TIMEOUT: -1

I'm checkout the documentation but couldn't find any which is defaulted to 60s and has impact for this command. I'm missing something? I report this as bug because there seems to be no documentation for this issue to be resolved and set a higher timeout.

Gitea Version

1.22.0

Can you reproduce the bug on the Gitea demo site?

Yes

Log Gist

No response

Screenshots

No response

Git Version

No response

Operating System

linux

How are you running Gitea?

Kubernetes with traefik ingress

Database

MSSQL

techknowlogick commented 4 days ago

Can you update to 1.22.1 (has just been tagged, and the blog post hasn't even been written yet), as it has some improvements around this. As well, are you using minio (or s3) to store packages?

CPlusPlus17 commented 3 days ago

I updated to 1.22.1, but the problem is still the same. We'are using selfhosted minio as storage. (Clean logs)

2024/07/05 06:39:24 ...eb/routing/logger.go:68:func1() [W] router: slow      PATCH /v2/xxx/dotnet/build/blobs/uploads/ht5ptpwgw1drkjmxh0gcpxwox for 10.244.2.206:56532, elapsed 3123.9ms @ packages/api.go:700(packages.ContainerRoutes.func2.2)
2024/07/05 06:40:21 ...ntainer/container.go:90:apiError() [E] unexpected EOF
2024/07/05 06:40:21 ...eb/routing/logger.go:102:func1() [I] router: completed PATCH /v2/xxx/dotnet/build/blobs/uploads/ht5ptpwgw1drkjmxh0gcpxwox for 10.244.2.206:56532, 500 Internal Server Error in 60199.3ms @ packages/api.go:700(packages.ContainerRoutes.func2.2)