-
Bug
-
Resolution: Done-Errata
-
Major
-
6.16.0
Description of problem:
When updating Satellite\Capsule 6.16 from snap 3 to 5, if the update attempt fails in between e.g. during installer execution, Despite fixing the issue and re-runing the update, All the steps start from scratch instead of skipping them and simply starting with the installer part.
How reproducible:
Always
Is this issue a regression from an earlier version:
Yes
Steps to Reproduce:
1. Install a Satellite\Capsule 6.16 from older snap ( snap 1 or 2 )
2. Stage the required repos on the filesystem to update their versions to Snap 5
3. In /etc/hosts create a bad FQDN entry of the main IP address , so that DNS lookup of IP fails to match the expected FQDN
4. Use "satellite-maintain update run" to perform the update along with necessary whitelists as needed.
5. Fix /etc/hosts
6. Redo step 4.
Actual behavior:
Step 4: During update run , all checks completed, services are stopped , packages are updated and then installer execution started but fails since the DNS lookup is pointing to the bad FQDN
Step 6. During re-execution of update run, it asks me to start the services first as all the checks are happening again and then basically it does everything as it did exactly before instead of skipping all the completed steps and simply resume from installer step.
Expected behavior:
Wherever possible, in case of failure during update, the re-execution of satellite-maintain should not redo all the checks and everything but simply resume from the failed step and skip the pre-checks.
Business Impact / Additional info:
- It could be applicable for major version upgrade as well but for now, i only was able to test for update .
- Increased downtime for update\upgrade and the all `do-over` stuff may even technically break something that we are not even aware of yet.
- links to
-
RHBA-2024:140284 Important: Satellite 6.16.0 release