[SERVER-49949] Reconstructing prepared transactions containing multi-key writes crashes the initial syncing node. - MongoDB Jira

XML

Word

Printable

JSON

Details

Type: Bug
Status: Closed
Priority: Major - P3
Resolution: Fixed
Affects Version/s: None
Fix Version/s: 4.4.1, 4.7.0
Component/s: Replication, Storage
Labels:
None

Backwards Compatibility:
Fully Compatible
Operating System:
ALL
Backport Requested:

v4.4, v4.2
Sprint:
Repl 2020-08-24
Linked BF Score:
20

Description

On-disk catalog multi-key update for multi-statement transactions happens in a side transaction block. ~~SERVER-48010~~ replaced ghost timestamp (determined by logical clock) with no-op write for timestamping the on-disk catalog writes on primaries. Since initial sync doesn't set transaction recoveryPrepareOpTime while reconstructing prepared transactions, it wrongly calls primary no-op write code path to acquire timestamp for the catalog write. As a result, we hit this invariant failure which leads to initial syncing node server crash.

This happens in both 4.4 and master.

Attachments

Issue Links

is related to

SERVER-53932 Multikey write during recovery of prepared transaction could use commit timestamp < stable timestamp

Closed

related to

SERVER-42251 Cannot timestamp multikey write during recovery of prepared transaction before LogicalClock has been initialized

Closed

SERVER-48010 Substitute ghost timestamp with no-op write in multi-statement txn multikey sidetxn write

Closed

Activity

People

Assignee:: Lingzhi Deng

Reporter:: Suganthi Mani

Participants:: Githook User, Lingzhi Deng, Samyukta Lanka, Suganthi Mani, Tess Avitabile

Votes:: 0 Vote for this issue

Watchers:: 10 Start watching this issue

Dates

Created:: Jul 28 2020 04:15:40 PM UTC

Updated:: Jan 20 2021 10:36:25 PM UTC

Resolved:: Aug 14 2020 05:25:39 PM UTC