Loading...

XML

Word

Printable

JSON

Type: Task
Resolution: Unresolved
Priority: Major - P3
Fix Version/s: None
Affects Version/s: None
Component/s: None
Labels:
None

Assigned Teams:

Replication
CAR Domain/s:
None

Aha! Reference:
None
Tracking Level:
None
Risk Status:
None
Exec Notes:
None
Goal Name(s):
None
Goal Link:
None

Scheduling tasks on a cancellation token will keep those tasks alive in-memory until the token's underlying CancellationSource is destructed. If a cancellation source stays alive for a long time and has many tasks scheduled on it, this can lead to significant memory use, as in ~~SERVER-92236~~. We should audit our use of cancellation sources for any that can live for a long time and probably rewrite them to at least periodically be cleared.

depends on

SERVER-112088 Audit use of long lived CancellationSources in Replication code

Needs Scheduling

SERVER-112089 Audit use of long lived CancellationSources in Cluster Scalability code

Needs Scheduling

SERVER-112090 Audit use of long lived CancellationSources in N&O code

Needs Scheduling

SERVER-112091 Audit use of long lived CancellationSources in QE code

Needs Scheduling

SERVER-112092 Audit use of long lived CancellationSources in CAR code

Needs Scheduling

is related to

SERVER-103841 Memory leak in TransactionCoordinator associated to long-lived cancellation source

Closed

SERVER-103945 Better memory management for CancellationToken

Backlog

related to

SERVER-92236 Chunk migrations should use short lived cancellation sources

Closed

SERVER-107121 migration_batch_inserter.cpp should use cancellation tokens instead of checking for interrupt on operation contexts

Backlog

(2 is related to, 2 related to)

Assignee:: Unassigned
Reporter:: Jack Mulrow
Participants:: Jack Mulrow
Votes:: 0 Vote for this issue
Watchers:: 17 Start watching this issue

Created:: Jul 11 2024 02:58:40 PM UTC
Updated:: Oct 07 2025 06:06:47 PM UTC

Details

Description

Attachments

Issue Links

Activity

People

Dates