Hello,
In our environment we are recreating host polls every month, to refresh hosts and avoid monthly pathing - as new image contains latest patches.
Currently re-image host pool is configured to be triggered out of business hours, when impact for users is low.
ISSUE: Each re-imaging introduces the risk, azure is not working properly or any other components do not work properly - this can lead to situation rebuilding is failing, or even all host pool is rebuilder but contains errors.
We will know about the issue next business day - and there is no easy roll back.
SOLUTION:
Introduce new re-imaga workflow
- double number of session host in host pool, and set all old hosts host in new state: exclude from scale OUT (separate request created for that), alternatively there should be option to set new hosts in new state: exclude from scale OUT
- in case of issues with new hosts rollback is easy, activating old session hosts, alternatively somebody confirms new session host are ok, and activate those
- after configurable time, old session hosts are deleted, and number of session hosts are set to original number.
SUMMARY
This alternate re-image workflow will reduce risk of rebuilding host pools with huge number of users, where any issue brings huge impact for business.
Best regards
Marek
Comments (1 comment)