Closed snorberhuis closed 5 months ago
Thanks for raising I'll investigate. If the problem is in the upstream library, I'll figure out some mitigation. whilst the dependancy is fixed. switching from delete to disable and retrying on empty datasets would be worth considering.
Out of interest how many large users/groups in you syncing?
I'm wondering whether the slow response might due to assembling the response with many entries. It might help me build a test harness.
I am syncing 1 user and 5 groups.
Thank you for fixing this issue @ChrisPates !
No problem, I have a couple of other bug fixes to complete and then I'll push a new release to the serverless application repository.
Describe the bug AWS SSO Sync running in a cron cycle detects no users or groups and removes all the users and groups in AWS. The next cycle, it will detect all users and groups and create them again. Permission Set assignments are lost resulting in breaking access for engineers.
This was also reported here:https://github.com/awslabs/ssosync/issues/159#issuecomment-1927312096
For me it seems likely a HTTP call is slow or unresponsive somewhere on the Google and this failure is not correctly caught resulting in assuming users and groups are deleted.
Logs before deletion:
Logs during deletion cycle as confirmed by CloudTrail
Next cycle where users are created
To Reproduce Steps to reproduce the behavior: Unknown
Expected behavior Users and groups are created once.
Additional context The sync has been running continually since December