Details
-
Bug
-
Resolution: Fixed
-
Critical
-
7.2.4
-
Untriaged
-
0
-
Unknown
Description
Lost replicas recreated for indexes in planning should be removed if there are no enough nodes to accomodate it.
There is corner case when recreated replicas will not be removed and will reside on old temp node in plan.
When this issue will occur if:
- Source cluster used to take backup has indexes with some lost replicas. (For index numReplica+1 > actual instances)
Thus repair will happen as part of restore. - Those multiple indexes for a bukcet have same <idx_name partitionId> combination,
ex:- Primary indexes without partitions on different scope/collection in bucket
<#primary 0> on b1.s1.c1
<#primary 0> b1.s1.c2
<#primary 0> b1.s1.c3
Attachments
Issue Links
- is a backport of
-
MB-62093 Planner not placing few indexes on cluster while restore
- Resolved