- ARCHIVED: Uyuni Users

IMPORTANT: Patch for Uyuni Server 2022.10
by Julio Gonzalez 14 Nov '22

14 Nov '22

Hello! ################################################# WARNING: Instead of running `zypper up` as provided by the instructions at the website, we strongly recommend using `zypper up|tee /tmp/update.log` or similar, so you can find errors during the migration in the unlikely event they happen ################################################# Several users at Gitter and GitHub have reported that after the update to Uyuni 2022.10, and after applying the patch from last week, Cobbler had problems to start because templates files were not found, so some features were broken, such as: - Onboarding new clients - Removing existing clients - Autoinstallation In such cases you can see errors in Cobbler like: cobbler.cexceptions.CX: 'serializer: error loading collection profile: Invalid automatic installation template file location /var/lib/cobbler/templates/ xxxxxxxxxxxx.cfg, file not found This problem was caused because the "autoinstall" attribute for the Cobbler collections was wrongly migrated at the time of running the migration from the old Cobbler collections from version 2 to version 3, making Cobbler not able to find the "templates" files for the profiles anymore. Any subdirectory was removed from the "autoinstall" attribute. For "Autoinstallation" profiles generated using Uyuni, the "autoinstall" files are generated and placed by Uyuni at "upload/" and "wizard/" subdirectories of `/var/lib/cobbler/templates/`. These Cobbler collections and templates files (for Uyuni created profiles) are recreated by Uyuni in case they are missing, on every spacewalk-services restart. With this new upgrade, we have fixed the migration of the "autoinstall" attribute from Cobbler collections, and make the migration to run again in order to fix any possible inconsistency created by the execution of the wrong migration from the previous patch, so Cobbler will be now able to find the "autoinstall" templates and finish the startup. The migration will run automatically during the upgrade, if there is any unexpected issue during the migration you should see some errors when zypper is updating "cobbler" or "spacewalk-setup" packages. In the unlikely event there are still some conflicts for the `.cfg` files, you will get instructions provided by the upgrade, to fix the issue. A patch for Uyuni 2022.10 is now available, in case you are affected, so you do not need to wait for Uyuni 2022.11 (this version will also include this fix). You can apply the patch at the Uyuni Server by following the instructions at: https://www.uyuni-project.org/pages/patches.html (please be aware of the WARNING at the beginning of this email!) Thanks to all users that reported the issues, and helped with the debugging. Happy hacking! -- Julio González Gil Release Engineer, SUSE Manager and Uyuni jgonzalez(a)suse.com

1 0

IMPORTANT: Security patches and feature patches for Uyuni Server 2022.10
by Julio Gonzalez 04 Nov '22

04 Nov '22

Hello! ### Cobbler migration Some users have reported that after the Update to Uyuni 2022.10, there are still some problems related to Cobbler, where collections are not migrated, so the cobbler service fails to start at the Uyuni Server. In such situations some features can be broken, such as: - Onboarding new clients - Removing existing clients - XML-RPC via HTTP - spacecmd - spacewalk-common-channels This is not happening for most users, but we are now releasing a patch to address it, and migrate any collections provided by any past cobbler version. ### File descriptors leakage Some users reported that, after updating to Uyuni 2022.10, and after some days, the Server stopped working, and they saw a steady increase of "open files" or "allocated file descriptors". This was caused after the migration to log4j2. Every time a taskomatic job object gets created, it initializes its own logger programmatically to write its logs into a file. The lifecycle of a job object is relatively short, but the log4j2 runtime doesn't know it. When it gets permissions to write into a file, there's nothing to release it after the job is over. Moreover the log4j2 keeps track of loggers based on the class names. It effectively means we mutate the same logger object in memory over and over again with every new taskomatic job. The fix adds functionality to re-initialize the logger when a new job comes around, dropping any file handler it has. ### Security fixes During a SUSE Manager and Uyuni security audit, the SUSE security team reported three vulnerabilities at the spacewalk-java source package (which provides spacewalk-java-* or spacewalk-taskomatic packages): * CVE-2022-31255: Fix directory path traversal vulnerability * CVE-2022-43754: Fix reflected cross site scripting vulnerability * CVE-2022-43753: Fix arbitrary file disclosure vulnerability You can find more details at: https://www.suse.com/security/cve/CVE-2022-31255.html https://www.suse.com/security/cve/CVE-2022-43754.html https://www.suse.com/security/cve/CVE-2022-43753.html A patch for Uyuni 2022.10 to fix all the issues was just published, and because of the security issues, it is strongly recommended you update your server, even if you are not affected by the cobbler migration issues or the file descriptors leakage. You can apply the patch at the Uyuni Server by following the instructions at: https://www.uyuni-project.org/pages/patches.html Thanks to all users that reported the issues, and helped with the debugging. Happy hacking!

1 0

Query about config management best practice.
by Simon Avery 04 Nov '22

04 Nov '22

Hello, I've got a query about best practice for config management. We have several Configuration Channels, each with a bunch of files in. Machines are subscribed to appropriate channels based on role and distro. We want to push those files out daily to subscribed machines, replacing any that have been edited on-server. We currently achieve this with a bash script on the Uyuni vm, called by crontab overnight, using "spacecmd -y -- configchannel_forcedeploy channelname" This works fine, but I'm wondering if that's the best way, as it's configured and monitored only outside of the webui? Forcing a regular high state appears to push those files, plus updates the .repo files and generally getting things as Uyuni thinks they should be. The recurring schedule seems to be designed to do this. But I can't see a way to automatically set a recurring schedule? Whilst I can set one for existing machines using SSM, I can't see a way to add one via Activation Key nor automatically enroll all machines into setting a highstate once a day or so. How do others approach this? Is there a gui way, or should I stick with the spacecmd route? S

1 0

Uyuni webui stops responding.
by Simon Avery 27 Oct '22

27 Oct '22

Hi Has anyone encountered an issue with Uyuni where the web interface becomes unresponsive? I've never had this happen before in at least a couple of years usage, but over the past couple of weeks we've had it happen several times to different users and when it happens, it affects all users. The user's web browsers report that no data is sent on page link or reload. Manually running "spacewalk-service restart" brings it back into life within a couple of minutes. This isn't related to the 2022.10 update, since I've held back applying that so far and we're still on 2022.08. The only thing that's changed is a couple of new users, and perhaps something is taking exception to them. (But we're all on similar chrome + windows clients) I've not yet done any serious debugging (each time the priority has been to get it back up), but would appreciate any tips from anyone that's encountered this. (I'll try and remember to run "spacewalk-service status" next time to at least identify which service is faulting) Thanks Simon

3 4

clients can't connect to uyuni server, JWT token
by Haupt, Torsten 25 Oct '22

25 Oct '22

Hi, since some time my clients can't contact the uyuni server. A "zypper ref" shows for example: zypper ref Permission to access 'https://xxxuyuni.xxx.de:443/rhn/manager/download/sle12_5-utilities-x86_64/r…' denied. Abort, retry, ignore? [a/r/i/...? shows all options] (a): i Last contact time in uyuni is also a long time ago. If I copy the url above in a browser, it get: Token is not valid to access repomd.xml in sle12_5-utilities-x86_64: JWT (claims->{"exp":1652861921,"iat":1621325921,"nbf":1621325801,"jti":"kufB_AaMdzSgIr92uq9cfA","org":1,"onlyChannels":["sle12_5-utilities-x86_64"]}) rejected due to invalid claims. Additional details: [The JWT is no longer valid - the evaluation time NumericDate{1666073380 -> 18.10.2022, 08:09:40 MESZ} is on or after the Expiration Time (exp=NumericDate{1652861921 -> 18.05.2022, 10:18:41 MESZ}) claim value.] It seems a token is to old; but how to renew it? Greets Torsten

4 5

Canceled: Uyuni Community Hours
by Julio Gonzalez 24 Oct '22

24 Oct '22

Dear Community: This Friday most of the SUSE Manager team will not be available, including the moderators, so we need to cancel this session. We'll see you in the November session, and meanwhile remember the Community is available at the mailing lists, Gitter and GitHub. Sorry about the inconvenience!

1 0

IMPORTANT: Patch for Uyuni Server 2022.10
by Julio Gonzalez 24 Oct '22

24 Oct '22

Hello! Several users at Gitter have reported that after the Update to Uyuni 2022.10, and after applying the patch from last week, that the cobbler package fails to update, or that the cobbler service is not starting at the Uyuni Server. In such situations some features can be broken, such as: - Onboarding new clients - Removing existing clients - XML-RPC via HTTP - spacecmd - spacewalk-common-channels The problem was causes because of bugs at migration scripts for cobbler itself. Such bugs are only happening for long time Uyuni users where the cobbler configuration is old enough, so most users should not notice this problem. If you are affected, you will see the following error while installing the cobbler package: > schema.SchemaError: Key 'proxy_url_ext' error: > None should be instance of 'str' Or the following error at the logs, when starting cobbler: > cobblerd[32057]: schema.SchemaError: Key 'allow_duplicate_hostnames' error: > cobblerd[32057]: 0 should be instance of 'bool' The fix is at the SR: https://build.opensuse.org/package/rdiff/systemsmanagement:Uyuni:Master/cob… (an OBS account is required to see the diff) A patch for Uyuni 2022.10 to fix both problems was just published, in case you are affected, and so you do not need to wait for Uyuni 2022.11 (this version will also include this fix) You can apply the patch at the Uyuni Server by following the instructions at: https://www.uyuni-project.org/pages/patches.html Thank to all users that reported the issue, and helped with the debugging Happy hacking! -- Julio González Gil Release Engineer, SUSE Manager and Uyuni jgonzalez(a)suse.com

1 0

Disconnected Setup with SMT results in "java.io.FileNotFoundException: /path/to/product_tree.json"
by Srinidhi B 20 Oct '22

20 Oct '22

Hello List, I'm trying to use the contents of my existing SMT12 setup in Uyuni and I've followed the steps documented at: https://www.uyuni-project.org/uyuni-docs/en/uyuni/administration/disconnect… But I'm getting the following error when I run "mgr-sync refresh": uyuni:~ # mgr-sync refresh Refreshing Channel families Please enter the credentials of SUSE Manager Administrator. Login: admin Password: [DONE] Refreshing SUSE products [FAIL] Error: <Fault -1: 'redstone.xmlrpc.XmlRpcFault: unhandled internal exception: com.suse.scc.client.SCCClientException: java.io.FileNotFoundException: /media/SMT/repo/disconnected-setup/product_tree.json (No such file or directory)'> Detailed steps: =============== We use a NFS mount for our SMT servers. On the same NFS mount, I created a new sub-directory called "disconnected-setup" and ran the following two commands: # smt-sync --createdbreplacementfile /tmp/dbrepl.xml # smt-mirror --dbreplfile /tmp/dbrepl.xml --directory /srv/www/htdocs/repo/disconnected-setup/ --fromlocalsmt -L /var/log/smt/smt-mirror-export.log # ls -l /srv/www/htdocs/repo/disconnected-setup/ total 11848 -rw-r--r-- 1 smt www 2 Oct 3 00:07 organizations_orders.json -rw-r--r-- 1 smt www 10006809 Oct 3 00:07 organizations_products_unscoped.json -rw-r--r-- 1 smt www 1744853 Oct 3 00:07 organizations_repositories.json -rw-r--r-- 1 smt www 341708 Oct 3 00:07 organizations_subscriptions.json drwxr-xr-x 4 smt www 4096 Oct 3 04:28 repo lrwxrwxrwx 1 smt www 9 Oct 3 04:28 SUSE -> repo/SUSE Back on the Uyuni server, I ran: # mount -t nfs storage:/srv/ftp/SMT /media/SMT # echo 'server.susemanager.fromdir = /media/SMT/repo/disconnected-setup/' >> /etc/rhn/rhn.conf # systemctl restart tomcat Could someone please tell me what am I missing? How do I get the "product_tree.json" file generated from the SMT server? Regards, Srinidhi.

2 2

IMPORTANT: Patch for Uyuni Server 2022.10
by Julio Gonzalez 20 Oct '22

20 Oct '22

Hello! Several users at GitHub, mailing lists and Gitter have reported that after the Update to Uyuni 2022.10, cobbler is not starting at the Uyuni Server, therefore affecting some features, such as: - Onboarding new clients - Removing existing clients - XML-RPC via HTTP - spacecmd - spacewalk-common-channels The following bug reports are related to this problem: - https://github.com/uyuni-project/uyuni/issues/6037 - https://github.com/uyuni-project/uyuni/issues/6041 - https://github.com/uyuni-project/uyuni/issues/6045 The problem was caused for a bug at the script for the migration of the settings and collections to Cobbler 3.3.3 The fix is at the PR https://github.com/uyuni-project/uyuni/pull/6050 A patch for Uyuni 2022.10 to fix both problems was just published, in case you are affected, and so you do not need to wait for Uyuni 2022.11 (this version will also include this fix) You can apply the patch at the Uyuni Server by following the instructions at: https://www.uyuni-project.org/pages/patches.html Thank to all users that reported the issue, and helped with the debugging Happy hacking! -- Julio González Gil Release Engineer, SUSE Manager and Uyuni jgonzalez(a)suse.com

1 0

no more OES2018 repos in Uyuni after new key for SLES subscriptions
by Thomas Weis 15 Oct '22

15 Oct '22

Hello everyone we synchronize the current OES2018-SP3 repositories via Uyuni. After a maintenance renewal, we received a new key for the SLES subscription because our current subscription could no longer be activated with the existing key. With the new key we can now synchronize all SLES repos within the scope of the subscription again, but obviously all OES2018 repos are missing. I am sure that with the old key all OES2018 repos were listed and could be selected. Are we doing something wrong or does something have to be activated in addition to the new subscription key? Thanks for the support! Thomas Mit freundlichen Grüßen / Kind regards Thomas Weis NTW Netzwerk • Telekommunikation • Datentechnik Burghaldenstrasse 24 D-71336 Waiblingen

2 3