Yum repos failing one by one on a previously working NXRM 3 server

Sonatype Nexus Repository
1

I previously had NXRM 3 setup to proxy all the CentOS7 repos I needed for a server. The repos I had setup were for base, epel, epel-debuginfo, extras, centosplus, and updates. First epel-debuginfo failed with java.net.uknownhostexception, even though its the same host as every other repo. I removed this repo and now others are failing with 502 Bad Gateway from the CentOS client, or Error: requested datatype primary not available.

Doing a yum clean all doesn’t help as I’ve done this and a yum makecache fails with the datatype primary error on different repos. If I comment one out, it’ll just move to another, in alphabetical order.

How do I know what repo is causing the error when it changes every time I do a yum command?

NXRM3 server:
CentOS 7.6.1810
Nexus 3.17.0-01

Client is CentOS 7.6.1810 as well

thanks,
James

# rm -rf /var/cache/yum
# yum clean all
Loaded plugins: fastestmirror
Cleaning repos: base epel extras grafana updates
# yum makecache
Loaded plugins: fastestmirror
Determining fastest mirrors
base                                                                                             | 2.2 kB  00:00:00
epel                                                                                             |  318 B  00:00:00
extras                                                                                           | 1.9 kB  00:00:00
grafana/signature                                                                                |  488 B  00:00:00
grafana/signature                                                                                | 2.9 kB  00:00:00 !!!
updates                                                                                          | 1.9 kB  00:00:00
(1/15): base/group_gz                                                                            | 166 kB  00:00:00
(2/15): base/primary                                                                             | 2.8 MB  00:00:00
(3/15): base/other                                                                               | 1.6 MB  00:00:00
(4/15): extras/prestodelta                                                                       |  65 kB  00:00:00
(5/15): extras/filelists                                                                         | 215 kB  00:00:00
(6/15): extras/primary                                                                           |  92 kB  00:00:00
(7/15): extras/other                                                                             |  92 kB  00:00:00
(8/15): base/filelists                                                                           | 7.2 MB  00:00:00
(9/15): grafana/primary_db                                                                       |  41 kB  00:00:00
(10/15): grafana/filelists_db                                                                    | 679 kB  00:00:00
(11/15): grafana/other_db                                                                        | 8.5 kB  00:00:00
(12/15): updates/prestodelta                                                                     | 857 kB  00:00:00
(13/15): updates/primary                                                                         | 3.6 MB  00:00:00
(14/15): updates/other                                                                           | 549 kB  00:00:00
(15/15): updates/filelists                                                                       | 5.9 MB  00:00:02
base                                                                                                        10019/10019
base                                                                                                        10019/10019
base                                                                                                        10019/10019
Error: requested datatype primary not available
2

Check the request.log to see if the 502’s are coming from Nexus. If they are not, debug whatever reverse proxy is running in Nexus. If they are, then try enabling outbound request logging to get more information:

3

The nexus log on the console shows 304s for outbound requests to centos.org. The request.log shows 200s and by the output I gave yesterday, it pulls these cached objects just fine.

I’ll go through the repos one by one to figure out the Error: requested datatype primary not available. It looks like the 502s stopped last night, maybe a network error?

Thanks,
James

4

For some reason the fedora backed mirrors (epel, epel-debuginfo) are causing that issue.

Time to figure out what the deal is there. Thanks for your assistance.