Submitted by agathongroup on Mon, 07/24/2017 - 23:41
We've been getting errors like the following in our cron.hourly jobs (running yum):
Not using downloaded repomd.xml because it is older than what we have:
Current : Thu Jul 20 22:38:19 2017
Downloaded: Thu Jul 20 22:34:25 2017
Not using downloaded repomd.xml because it is older than what we have:
Current : Thu Jul 20 22:38:17 2017
Downloaded: Thu Jul 20 22:34:23 2017
We traced this back to the Virtualmin repomd.xml; note the timestamp:
[root@wotw ~]# ll /var/cache/yum/x86_64/7/virtualmin/repomd.xml
-rw-r--r-- 1 root root 2519 Jul 20 22:38 /var/cache/yum/x86_64/7/virtualmin/repomd.xml
My hunch is that of the two servers behind software.virtualmin.com (for the yum repo), one of them has a time that's off, possibly by about four minutes.
Could someone please check and confirm that NTP is configured, running, and currently working on both servers behind software.virtualmin.com?
Thanks! Peter
Status:
Closed (fixed)
Comments
Your hunch is pretty close to right, and should be fixed now.
The problem is we added a mirror for software.virtualmin.com due to a hardware failure a couple weeks ago. Problem is, the way I push software into the yum repos is built around one server. And, for a variety of historic reasons, simply doing an "rsync" of one to the other wasn't desirable. So, repo metadata was being generated on each mirror...and it led to this error and one other weirder one. But, now we generate repo metadata in one place and sync it to all mirrors, and the old mirror has been taken offline for good.
See this forum post for more detail: https://www.virtualmin.com/node/52916
Do a
yum clean all
on any affected hosts to make this error go away and pick up new metadata.Submitted by agathongroup on Tue, 07/25/2017 - 09:27 Comment #3
Fantastic, thanks for the quick and thorough response and fix!
Submitted by IssueBot on Tue, 08/08/2017 - 09:30 Comment #4
Automatically closed - issue fixed for 2 weeks with no activity.