[SERVER-30667] Recover to a Timestamp Up/Down Testing: Successful Upgrade/Downgrade Scenario Created: 15/Aug/17  Updated: 06/Dec/22  Resolved: 06/Apr/18

Status: Closed
Project: Core Server
Component/s: Replication, Storage
Affects Version/s: None
Fix Version/s: None

Type: Task Priority: Major - P3
Reporter: Cristopher Stauffer Assignee: Backlog - Replication Team
Resolution: Duplicate Votes: 0
Labels: rollback-optional
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Depends
depends on SERVER-30349 Atomically turn on recovery to a time... Closed
Duplicate
duplicates SERVER-32590 RTT 3.6<->4.0 upgrade/downgrade. Closed
Assigned Teams:
Replication
Participants:

 Description   

Scenarios:
Variants: Cluster, Standalone

As an administrator with a 3.4 mongod instance,
When I shutdown the instance and start under the 3.6 binary,
then the instance starts successfully*.

  • For standalone instances, iterating over all collection statuses should yield log=true
  • For replica cluster, iterating over all collection statuses should yield log=true for all user collections and no otherwise.

As an administrator with a 3.6 mongod instance and FCV 3.4,
When I shutdown the instance and start under the 3.4 binary,
then the instance starts successfully*.

  • For standalone instances, iterating over all collection statuses should yield log=true
  • For replica cluster, iterating over all collection statuses should yield log=true.


 Comments   
Comment by Cristopher Stauffer [ 29/Aug/17 ]

I agree about your comment on standalone. I'll update it. This is intended as a future test for the multi-version test suite.

Comment by Susan LoVerso [ 29/Aug/17 ]

cristopher.stauffer I'm not sure if this is a future unit test outline or something else.

Let me elaborate on what you've written for clarification. First I suggest the term standalone instead of single instance because that phrase can be confused with single node replica set.

When running a 3.6 binary (regardless of FCV value):

  • Running a standalone, all collection statuses should yield log=true.
  • Running with replication, all replicated user collection statuses should yield log=false. The oplog and user tables in the local database should yield log=true.

If you perform a clean shutdown of a 3.6 binary and FCV 3.4 and restart with a 3.4 binary your statements are correct that all collection statuses should yield log=true.

Generated at Thu Feb 08 04:24:36 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.