Problem:
I am trying to add a new manifest but having conflict with the old one since I removed the old manifest file from redhat portal Expected outcome:
Foreman should upload the newly dowloaded manifest Foreman and Proxy versions:
|2.0.3| Foreman and Proxy plugin versions:
Distribution and version:
Other relevant data:
From the image attached, the Manifest_CAO_Development is still appearing after deleting manifest and removing cache. If we try to upload the new one, the result is stuck at 0% importing.
Hi @almond. If you go to Monitor → Tasks, you should see that task near the top of the list (Manifest Import). If you click in to that task you can see more details about the individual steps on that task. And, by clicking directly on those individual tasks you can see even more specific information. Do you see any failed (red) tasks in the Plan or Run sections in the Manifest Import task details? Please share those details here so we can get an idea of what’s going on.
It also sounds like the first Manifest Delete task that you tried did not succeed. Definitely look into the details of that task as well.
Ok, now we’re getting somewhere. Would it be possible for you to share your /var/log/candlepin/candlepin.log with me? It will be too large to paste here, but perhaps something like Dropbox would work. And if you like you can message me privately with that log file.
My hunch is that since we unknowingly deleted the subscription manifest from redhat portal first rather than deleting it from the foreman server, may be it might be reaching out to the old url. So it must be some db error.
We support running Foreman+Katello in environments that are disconnected from the Internet so there shouldn’t be any issue with the scenario you described. The new manifest should get imported. Are you seeing the same problem in the new manifest import task? And can you upload the new /var/log/candlepin/error.log so I can confirm the error has not changed on the Candlepin side?
@Jonathon_Turel Tried everything. Imported new manifest, went inside postgres db and tried to change inside candlepin database. But that also failed. We are out of ideas now. Happy to have a call if needed. I’ve sent my email ID as direct message.
I hopped on a call with @almond and we found that the root cause went back to the deletion of the manifest on the Katello side. The task reported success but looking in Dynflow showed that deletion was actually stuck by some previously failed job. /var/log/candlepin/error.log showed that the system disk was full according to Artemis which by default considers full to be 90% usage. After adding <max-disk-usage>99</max-disk-usage> to the candlepin broker.xml and clearing the cp_async_jobs table, the manifest delete succeeded which unblocked import of the new manifest!