[zmqorch]: Restore drain-until-empty consumer for non-DPU/fabric#4578
Merged
prsunny merged 3 commits intoMay 27, 2026
Merged
Conversation
Collaborator
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Contributor
Author
vivekrnv
reviewed
May 15, 2026
venkit-nexthop
force-pushed
the
fix-zmqorch-route-perf-regression
branch
from
May 15, 2026 19:28
c18add4 to
847130e
Compare
Collaborator
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
venkit-nexthop
force-pushed
the
fix-zmqorch-route-perf-regression
branch
from
May 15, 2026 19:34
847130e to
5b7b4b3
Compare
Collaborator
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
venkit-nexthop
force-pushed
the
fix-zmqorch-route-perf-regression
branch
from
May 15, 2026 19:41
5b7b4b3 to
a842a56
Compare
Collaborator
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
venkit-nexthop
force-pushed
the
fix-zmqorch-route-perf-regression
branch
from
May 15, 2026 20:32
a842a56 to
f8bf093
Compare
Collaborator
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Contributor
|
LGTM |
venkit-nexthop
force-pushed
the
fix-zmqorch-route-perf-regression
branch
from
May 16, 2026 16:06
f8bf093 to
489cad7
Compare
Collaborator
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
venkit-nexthop
force-pushed
the
fix-zmqorch-route-perf-regression
branch
from
May 17, 2026 04:25
489cad7 to
30fc27f
Compare
Collaborator
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
PR sonic-net#3910 changed ZmqConsumer::execute() to perform a single batch pop per invocation, primarily to reduce peak memory in DASH (DPU) workloads. For non-DPU / non-fabric switches that receive high volumes of APPL_DB route updates from fpmsyncd, this regressed throughput: each execute() call no longer drains the consumer table, so updates take many more polls to apply. Introduce a ZmqRouteConsumer subclass that restores the prior drain-until-empty loop in execute(). ZmqOrch::addConsumer() selects ZmqRouteConsumer for non-DPU / non-fabric switch types; DPU and fabric continue to use ZmqConsumer to preserve the bounded-memory behavior added in sonic-net#3910. Also define gMySwitchType in orchagent/p4orch/tests/test_main.cpp so p4orch_tests links once zmqorch.cpp references the global. Signed-off-by: Venkit Kasiviswanathan <venkit@nexthop.ai>
venkit-nexthop
force-pushed
the
fix-zmqorch-route-perf-regression
branch
from
May 18, 2026 16:33
30fc27f to
d63ed1c
Compare
Collaborator
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Signed-off-by: Venkit Kasiviswanathan <venkit@nexthop.ai>
Collaborator
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Contributor
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Collaborator
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Contributor
Author
|
@prabhataravind @qiluo-msft @prsunny @vivekrnv @deepak-singhal0408 |
vivekrnv
approved these changes
May 27, 2026
Contributor
Author
|
thanks @vivekrnv |
github-project-automation
Bot
moved this from In Progress
to Done
in SONiC Routing Dashboard
Contributor
|
@venkit-nexthop : @deepak-singhal0408 : this is reversing the changes done here: #3269 Have we verified we are not regressing the previous behavior ? |
Contributor
Author
@abdosi sorry for the late response. I am catching up after my leave. |
Yogapriya-cisco
pushed a commit
to prhoskot/sonic-swss
that referenced
this pull request
Jul 13, 2026
…ic-net#4578) * [zmqorch]: Restore drain-until-empty consumer for non-DPU/fabric PR sonic-net#3910 changed ZmqConsumer::execute() to perform a single batch pop per invocation, primarily to reduce peak memory in DASH (DPU) workloads. For non-DPU / non-fabric switches that receive high volumes of APPL_DB route updates from fpmsyncd, this regressed throughput: each execute() call no longer drains the consumer table, so updates take many more polls to apply. Introduce a ZmqRouteConsumer subclass that restores the prior drain-until-empty loop in execute(). ZmqOrch::addConsumer() selects ZmqRouteConsumer for non-DPU / non-fabric switch types; DPU and fabric continue to use ZmqConsumer to preserve the bounded-memory behavior added in sonic-net#3910. Also define gMySwitchType in orchagent/p4orch/tests/test_main.cpp so p4orch_tests links once zmqorch.cpp references the global. Signed-off-by: Venkit Kasiviswanathan <venkit@nexthop.ai> Signed-off-by: Yogapriya Mohankumar <ymohanku@cisco.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
7 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What I did
PR #3910 changed
ZmqConsumer::execute()to perform a single batch pop per invocation. This change introduces aZmqRouteConsumersubclass that restores the prior drain-until-empty loop inexecute().ZmqOrch::addConsumer()selectsZmqRouteConsumerfor non-DPU / non-fabric switch types, while DPU and fabric continue to useZmqConsumer.Also defines
gMySwitchTypeinorchagent/p4orch/tests/test_main.cppsop4orch_testslinks oncezmqorch.cppreferences the global. Production (main.cpp) and mock tests (mock_orchagent_main.cpp) already define it.Why I did it
PR #3910's single-batch pop was primarily motivated by reducing peak memory in DASH (DPU) workloads. For non-DPU / non-fabric switches receiving high volumes of APPL_DB route updates from fpmsyncd, the change regressed throughput: each
execute()call no longer drains the consumer table, so updates take many more polls to apply.This is a targeted temporary fix that preserves the bounded-memory behavior added in #3910 for DPU/fabric, while restoring the earlier drain semantics for the workloads that depend on them.
How I verified it
gMySwitchType: productionorchagentresolves viamain.cpp,mock_testsviamock_orchagent_main.cpp, andp4orch_testsvia the new definition intest_main.cpp.Details if related