#7403 Sometimes OpenShift builds hang while curling the fedora infra yum repo config
Closed: Fixed 5 years ago Opened 5 years ago by bowlofeggs.

  • Describe what you need us to do:
    I've got a Bodhi build in OpenShift that's been running for 38 minutes, most of it waiting on a curl command to finish:
oc logs -f builds/bodhi-web-16
Step 1/11 : FROM fedora:29
 ---> 8c568f104326
Step 2/11 : LABEL name "bodhi-web" vendor "Fedora Infrastructure" license "MIT"
 ---> Running in 62b50e962a7c
 ---> 019aa3aa81d5
Removing intermediate container 62b50e962a7c
Step 3/11 : RUN curl -o /etc/yum.repos.d/infra-tags.repo https://infrastructure.fedoraproject.org/cgit/ansible.git/plain/files/common/fedora-infra-tags.repo
 ---> Running in 4c4f420e2659

  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100   271  100   271    0     0   5113      0 --:--:-- --:--:-- --:--:--  5211

This same thing happened last week when I build bodhi-3.11.0 in OpenShift. I'm not sure what the problem is, but it seems to be repeatable at least once (meaning, this is the second time I've seen it). I think last time this happened, it eventually just finished and continued, though I can't remember for sure.

  • When do you need this? (YYYY/MM/DD)
    This week would be good, so I can get 3.11.1 out the door.

  • When is this no longer needed or useful? (YYYY/MM/DD)
    If the problem goes away, we stop using OpenShift, or we stop using the infra repo.

  • If we cannot complete your request, what is the impact?
    I may not be able to deploy Bodhi releases, or maybe it'll just take a long time.


The other really strange thing is that curl is reporting that the download is 100% finished, so what is it waiting on? Is the other end of the connection just not sending an EOF or something?

@smooge helped me today by cancelling the build for me. We tried changing the URL to use http (i.e., no TLS) and an IP address (i.e., no DNS), but ultimately it ended up working no matter what URL we tried (including the original URL) and we were not able to reproduce the problem.

I will close this for now. We can reopen if it happens again.

Metadata Update from @bowlofeggs:
- Issue close_status updated to: Fixed
- Issue status updated to: Closed (was: Open)

5 years ago

Login to comment on this ticket.

Metadata