Add Decomission logic for Scale-out extents

[jzilak]

Hello,

it would be nice to have implemented decommission logic for scale-out repository extents. Current feature Evacuate extent is too buggy and easily ruined storage performance and fill all available slots for even for days or weeks.

I will prefer some kind of flags, like "in decommission" and extend will operate normally but only in read-only mode, till all stored backups expires and was deleted by retention policy. After that that goal system will change flag to "ready to decommission" and extent can be easily remove from production.

josef

[Shestakov]

Hello Josef, thanks for the request, it`s taken into account!
Regarding evacuation operation, what product version are you at? What kind of bugs do you experience?
Thanks!

[jzilak]

Hello,

we use 9.5 Update 2. Evacuation problems:
1. limiting jobs for evacuated repository to 1 cause evacuation stop working and freeze jobs with data stored on that repository (looks like limit is hard even for read only operations and db transactions)
2. evacuation ReFS repository with no limits cause system unresponsiveness/crash due to extreme load to ReFS subsystem

josef