Rocky 8.10 rebuild hangs on large package update

I recently rebuilt a Rocky 8.10 host and it hung while downloading a large package during the Kickstart upgrade phase. Here is the error from the Anaconda console:

[MIRROR] linux-firmware-20250626-131.gitb05fabcd.el8_10.norarch.rpm: Curl error (18): Transferred a partial file for https://rocky-linux-us-east4.production.gcp.mirrors.ctrliq.cloud/pub/rocky//8.10/BaseOS/x86_64/os/Packages/l/linux-firmware-20250626-131.gitb05fabcd.el8_10.norarch.rpm [transfer closed with 152245720 bytes remaining to read]

This happened twice, on successive attempts. By disabling the update phase in my Kickstart template, I got the build to complete. When the system was up and running, ‘dnf’ listed the problem package as ‘linux-firmware-20250626-131.gitb05fabcd.el8_10.noarch.rpm’, size 529 MB. I was able to run ‘dnf update’ and the package downloaded and installed just fine. Very strange.

The curl failure happened after about 10 minutes each time, so I am thinking, standard timeout? But what could be the difference between curl in the Kickstart/Anaconda session, vs. curl after system boot in a standard ssh session? — memory perhaps? This firmware package is almost the last to download in the upgrade phase and (possibly) the largest.

-jm

Do you have an approximate time that this happened? Several of us have tried that package and were not able to replicate the issue.

GCP has network outages and issues that sometimes impact downloads of large packages, so if I konw the time stamp, I can go look to see if we had any alerts at that time.

Thanks!

According to my foreman production log, the failed downloads happened sometime between 17:15 and 18:30 EDT on August 27, 2025.

-jm

It can happen if the content length does not agree with the number of bytes transferred. Next time it happens, you could try using curl outside of dnf to see if you get a clean download.

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.