We're updating the issue view to help you get more done. 

Slow/disconnected Async replication targets that should consume only small amount of packets will cause redolog to grow fast with unrelated operations

Description

Reproduction:
1. in client 1 create Notify Container that listen to type A
2. in client 2 write type B in a loop
3. disconnect client 1

Attached the 2 clients.

Expected:
As we don't need to save any notifications for client 1, the redo log should be 0

Actual:
The redo log is growing.

Same differentiation that we should do for types, we need to do for operations as well. Means that when a client listen for write operation. We should not save in the redolog any operations beside write.

Workaround

None

Acceptance Test

None

Status

Assignee

Unassigned

Reporter

Yuval Dori

Labels

None

Priority

Medium

SalesForce Case ID

10581,11085, 11752,10877

Fix versions

None

Commitment Version/s

None

Due date

None

Product

None

Edition

None

Platform

All