Details
-
Task
-
Resolution: Won't Do
-
Major - P3
-
None
-
None
-
None
-
None
-
Storage Execution
Description
There's a few cases to consider here.
After moving the files from _restore into the dbpath, then an unclean shutdown during this point will leave files unowned by both mongod and WiredTiger. During startup recovery when aborting the restore procedure, these files need to be removed from the filesystem.
Table imports in WiredTiger are non-transactional. An unclean shutdown during this procedure could result in some files owned by WiredTiger but not known to mongod. During startup recovery when aborting the restore procedure, some tables may need to be dropped through WiredTiger.
Once the collections are imported in both the server and WiredTiger, the abort procedure will need to remove any durable catalog changes in the server as well as dropping tables in WiredTiger during startup recovery.
Attachments
Issue Links
- depends on
-
SERVER-62629 Create a new internal command that allows nodes to vote for committing or aborting the restore procedure
-
- Closed
-
-
SERVER-62630 Implement the first phase of the restore algorithm where the backup is imported into the server
-
- Closed
-
-
SERVER-62633 Implement the first phase of the restore algorithm where the oplog is replayed on restored collections
-
- Closed
-
-
SERVER-62634 Implement the final steps of the first phase restore algorithm
-
- Closed
-