search

davidalger / docker-images-php

PHP and PHP-FPM images for Docker based on EL 8
https://hub.docker.com/r/davidalger/php
MIT License
15 stars 16 forks source link

Issue with current image centos:8 build #12

Closed Luwdo closed 3 years ago

Luwdo commented 3 years ago

It looks like the maintainers of centos:8 images have some urgent updates causing epel mirror checksums to fail without the latest.

docker run -it --rm davidalger/php:7.4-fpm dnf update

This will fail at the moment with

Error: Failed to download metadata for repo 'epel': Cannot download repomd.xml: Cannot download repodata/repomd.xml: All mirrors were tried

Looks like there is a fix in latest 8 tag

docker run -it --rm centos:8 dnf update

If you could manually trigger your git workflow CI to get the update, much appreciated.

davidalger commented 3 years ago

@Luwdo They are rebuilding now: https://github.com/davidalger/docker-images-php/actions/runs/1309650119

mr-pmillz commented 3 years ago

After pulling the latest php:7.4-fpm release just now that the build trigger pushed, I am still getting the same error. 🤔 

❯ docker run -it --rm davidalger/php:7.4-fpm dnf update
CentOS Linux 8 - AppStream                                                                                                   5.1 MB/s | 9.3 MB     00:01
CentOS Linux 8 - BaseOS                                                                                                      4.1 MB/s | 7.5 MB     00:01
CentOS Linux 8 - Extras                                                                                                       19 kB/s |  10 kB     00:00
Extra Packages for Enterprise Linux Modular 8 - x86_64                                                                       1.4 MB/s | 956 kB     00:00
Extra Packages for Enterprise Linux 8 - x86_64                                                                               1.4 kB/s | 4.7 kB     00:03
Errors during downloading metadata for repository 'epel':
  - Downloading successful, but checksum doesn't match. Calculated: 9719bd3e7d8f3a217103b14eb122758d3d3a086a6a46eeeab953017a4a5907f9e96079b7faec9a7f836571713ce25d14e94abd8b7709a8717b78b8292221a621(sha512) 9719bd3e7d8f3a217103b14eb122758d3d3a086a6a46eeeab953017a4a5907f9e96079b7faec9a7f836571713ce25d14e94abd8b7709a8717b78b8292221a621(sha512)  Expected: d7866fd1604ef749ced5df995dfa050d491a4cf642c7b81797d5da38e6e80829758956deaec02405ab2dbcecff8cd6dcdb13b5559198101fd965daa1958e8500(sha512) be6e8f5d423e9a31310dd5a6450097c2a1253efeffd54ebcae0ded443e878feb851304ae77130f16ac021b27850c5ee83444f4e6650cad926684beb7648fc471(sha512)
Error: Failed to download metadata for repo 'epel': Cannot download repomd.xml: Cannot download repodata/repomd.xml: All mirrors were tried
Luwdo commented 3 years ago

Well now that is strange, I saw a recent update in centos:8 and assumed that was the issue they solved just from a quick sanity check with the base image.

I don't see anything in the Dockerfile's that would cause these images to fail checksums for the epel repo metadata for subsequent updates.

This is eerily similar to an issue we are seeing with nodesource where certain regions CDNs had updated metadata from akamai and they forgot to resign so you get checksum errors.

I should be able to pull this repo down and build locally.

Luwdo commented 3 years ago

Ok good, booted a ec2 to test build on an isolated network. Same error, I'll do some more testing see if I can narrow down whats doing it.

Luwdo commented 3 years ago

Well ok now, thanks for running the update on the images, after doing some testing. This issue has to be an epel-release repo bug.

docker run -it --rm centos:8 /bin/bash
dnf install -y epel-release
dnf install -y epel-release

The second install will cause the same checksum issue.

Luwdo commented 3 years ago

same issue with bypassing the mirrors

dnf install https://dl.fedoraproject.org/pub/epel/epel-release-latest-8.noarch.rpm
mr-pmillz commented 3 years ago

appears to be working now

Luwdo commented 3 years ago

Yep the maintainers of the epel RPM repo appears to have fixed their issue.