Red Hat Bugzilla – Bug 1271287
yum hourly cron errors even when configured not to update
Last modified: 2018-06-25 11:11:25 EDT
Created attachment 1082467 [details]
patch for bug
Description of problem:
I have yum-cron installed to do daily updates. Hourly conf is set not to download, update or output messages. However, sometimes when a different mirror gets selected, I start getting this message, every hour:
Not using downloaded repomd.xml because it is older than what we have
It also contains timestamp info. I get this message every hour until I run 'yum clean all' on the machine.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. run yum update
2. set hourly conf to 'no' for donwload, apply and message options
3. switch to mirror with older repo metadata
error/warning message every hour
Yum shouldn't even run to update metadate when it is not configured to download, update or output messages.
I've attached a patch that moved the update/download/message option check earlier in checkUpdates, and this issue goes away...
Good catch! To get this fix implemented now, though, the bug should be reported against yum itself: as yum-cron is now something that's built out of the main yum srpms rather than something I can fix.
so I've reassigned it to Jeff (he's getting the regular yum EPEL bugs). If this isn't correct, please let me know so we can get this patch to the right person.
Last time I looked, yum-cron was a RHEL package, built from yum.srpm. Thus this bug should be filed for Red Hat > Red Hat Enterprise Linux 7, component yum.
I changed the Product to RHEL 7, component yum.
Thank you for the patch. However, we decided not to apply it as it would break one particular scenario in which yum-cron is used to only fetch metadata. In such a scenario, both the download_updates and apply_updates options would be set to "no", and it's likely that update_messages would also be "no", which would cause yum-cron to not even fetch metadata when run. That is also the default configuration shipped with yum-cron (see /etc/yum/yum-cron-hourly.conf).
We will take a look at how this (if at all) can be properly fixed. For now, I'd suggest that you just disable the hourly yum-cron job if you don't need the metadata updates. With the patch, yum-cron would exit immediately (without doing anything) when used in that configuration anyway.
Oops, removing the accidental devack+
I am having a similar problem. All my servers (30+) produce various errors via email due to yum-cron-hourly.
Most of them look like:
Could not get metalink https://mirrors.fedoraproject.org/metalink?repo=epel-7&arch=x86_64 error was
14: HTTPS Error 503 - Service Unavailable
At first, I tried to modify the config file (/etc/yum/yum-cron-hourly.conf) but these errors are irrelevant to any of the config options, they are produced no matter what.
Eventually I had to disable yum-cron-hourly via my Ansible playbook, that was the only way that I could find to stop the errors.
A couple possible temporary workarounds to end the flood of 503 emails.
1. `# mv 0yum-hourly.cron /opt/0yum-hourly.cron.backup`
2. Filter them out in your mail client. If the emails go to a list of people, that's not ideal.
It's worth noting even setting the verbosity to -4 won't stop the 503 emails.
(In reply to Ray Foss from comment #11)
> It's worth noting even setting the verbosity to -4 won't stop the 503 emails.
This has been addressed upstream and a BZ has been created to backport this into RHEL: