SUSE Manager channel sync problems

SUSE Manager 4.1
SUSE Manager 4.0
SUSE Manager 3.1
SUSE Manager 3.0
SUSE Manager 2.1
SUSE Manager 1.7
SUSE Manager 1.2

This document is intended to list possible sync errors that may get encountered on SUSE Manager systems when running a channel sync either via mgr-ncc-sync or manually using the spacewalk-repo-sync command.

Errno 256 - No more mirrors to try

By default yum uses an internal 30 second timeout value to access / download files from a given repository. This timeout might get hit soon on busy servers. To increase the time please create /etc/rhn/spacewalk-repo-sync/yum.conf with the following content:

[main]
timeout=<value>

The <value> taken by timeout is in seconds, so timeout=300 would increase the yum timeout to 5 minutes which should allow enough time to finish the sync on a repository. Please monitor the behavior of the sync process and adjust the value if necessary. Also keep in mind that cloning channels from a repository where not all patches have been synced properly may cause problems to your test / staging / production environment.

See man 5 yum.conf or man spacewalk-repo-sync for more details.

Possible package remains in /var/spacewalk/packages/1/stage/

Another problem that causes a "Errno 256 - No more mirrors to try " are incomplete package downloads in /var/spacewalk/packages/1/stage/. If /etc//rhn/spacewalk-repo-sync/yum.conf was created but the issue is seen on specific channel syncs, please check the staging directory and delete any file that has been downloaded. If a cron job is in use to run repo sync periodically (see for details https://www.suse.com/documentation/suse_manager/singlehtml/book_susemanager_install/book_susemanager_install.html#s1-maintenance-auto-sync), please change the command from:

# connect to customer center every day at random time
# between 03:03 and 05:50
3 3 * * * sleep $[ $RANDOM / 5 ]; /usr/sbin/mgr-ncc-sync >/dev/null 2>/dev/null

to

# connect to customer center every day at random time
# between 03:03 and 05:50
3 3 * * * sleep $[ $RANDOM / 5 ]; rm /var/spacewalk/packages/1/stage/* && /usr/sbin/mgr-ncc-sync >/dev/null 2>/dev/null

Checksum mismatch

In any /var/log/rhn/reposync/*.log a message similar to the following appears:

Repo Sync Errors: (50, u'checksums did not match
326a904c2fbd7a0e20033c87fc84ebba6b24d937 vs
afd8c60d7908b2b0e2d95ad0b333920aea9892eb', 'Invalid information uploaded
to the server')
The package microcode_ctl-1.17-102.57.62.1.x86_64 which is referenced by
patch microcode_ctl-8413 was not found in the database. This patch has
been skipped.

Please run a manual repo sync using

spacewalk-repo-sync --channel <channelname> --type yum -Y

The -Y will run a deepverify on the repository data and not use locally cached checksums - this process may take some time depending on the size of the repository and its content.

Debugging connection problems

In case the above recommended settings do not solve the issue, please run:
 

export URLGRABBER_DEBUG=DEBUG
spacewalk-repo-sync -c <channelname> <options> > /var/log/spacewalk-repo-sync-$(date +%F-%R).log 2>&1

to check whether the communication encounters any problem.

In version 4.1 a lot of other operating systems gpg keys were included by default and such problems should disappear for openSUSE, CentOS, Oracle Linux, Ubuntu and Microfocus Open Enterprise Server.

The problem will still exist for channels such as the one providing Nvidia drivers, and maybe others.

Please see the full explanation in the release notes:
https://www.suse.com/releasenotes/x86_64/SUSE-MANAGER/4.1/#_third_party_gpg_keys_now_included