Error: Package: libkadm5-1.15.1-34.el7.x86_64 (base)

[wes hayutin]

http://logs.openstack.org/45/560445/238/check/tripleo-ci-centos-7-standalone/3a04cd5/job-output.txt.gz#_2019-02-04_00_14_51_012589

2019-02-04 00:14:51.012589 | primary | ---> Package pcre-devel.x86_64 0:8.32-17.el7 will be installed
2019-02-04 00:14:51.358653 | primary | --> Finished Dependency Resolution
2019-02-04 00:14:51.418117 | primary | Error: Package: libkadm5-1.15.1-34.el7.x86_64 (base)
2019-02-04 00:14:51.418421 | primary | Requires: krb5-libs(x86-64) = 1.15.1-34.el7
2019-02-04 00:14:51.418611 | primary | Installed: krb5-libs-1.15.1-37.el7_6.x86_64 (@updates)
2019-02-04 00:14:51.418756 | primary | krb5-libs(x86-64) = 1.15.1-37.el7_6
2019-02-04 00:14:51.418931 | primary | Available: krb5-libs-1.15.1-34.el7.x86_64 (base)
2019-02-04 00:14:51.419119 | primary | krb5-libs(x86-64) = 1.15.1-34.el7
2019-02-04 00:14:51.419274 | primary | Error: Package: krb5-devel-1.15.1-34.el7.x86_64 (base)
2019-02-04 00:14:51.419440 | primary | Requires: krb5-libs(x86-64) = 1.15.1-34.el7
2019-02-04 00:14:51.419612 | primary | Installed: krb5-libs-1.15.1-37.el7_6.x86_64 (@updates)
2019-02-04 00:14:51.419750 | primary | krb5-libs(x86-64) = 1.15.1-37.el7_6
2019-02-04 00:14:51.419906 | primary | Available: krb5-libs-1.15.1-34.el7.x86_64 (base)
2019-02-04 00:14:51.420043 | primary | krb5-libs(x86-64) = 1.15.1-34.el7

All upstream jobs are down on this issue atm.

[wes hayutin (weshayutin)]

testing change https://review.openstack.org/634622

[chandan kumar (chkumar246)]

Added the issue in centos bug tracker: https://bugs.centos.org/view.php?id=15775

[chandan kumar (chkumar246)]

From the logs, it appears that both libkadm5 and krb5 is in updates repo, might be landed at the same time but recently we are not seeing this in our CI.

[wes hayutin (weshayutin)]

Probably an issue w/ the upstream mirrors.

[Sorin Sbarnea (ssbarnea)]

This was caused by bad timing as the while the two repositories @updates and @base were updated.

If we would have follower the E405 rule of ansible-lint we would have avoided this failure.

Repositories are updated multiple time every day so the chance of hitting a similar issue is quite real.

Using a simple retry mecanism can make our build more resilient to networking issues.

See https://devops.stackexchange.com/questions/5829/how-to-solve-e405-remote-package-tasks-should-have-a-retry-in-ansible/5837

[Sorin Sbarnea (ssbarnea)]

In this specific case is not an ansible task but the same retry logic can be implemented on the bash function too.