Enable support for MQTT stitcher in stirling #1918
Conversation
Signed-off-by: Chinmay <[email protected]>
ChinmayaSharma-hue
temporarily deployed
to
pr-actions-approval
GitHub Actions
Inactive
Signed-off-by: Chinmay <[email protected]>
ChinmayaSharma-hue
temporarily deployed
to
pr-actions-approval
GitHub Actions
Inactive
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher.h
Outdated
Show resolved
Hide resolved
There was a problem hiding this comment.
@ChinmayaSharma-hue sorry for the late review and really appreciate all your work on the MQTT protocol support!
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher.cc
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher.cc
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher.cc
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher_test.cc
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher_test.cc
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher.cc
Outdated
Show resolved
Hide resolved
Signed-off-by: Chinmay <[email protected]>
ChinmayaSharma-hue
had a problem deploying
to
pr-actions-approval
GitHub Actions
Error
There was a problem hiding this comment.
Thanks for the latest updates! I know this latest round has quite a bit of comments, but stitching is a challenging component and the more thought that's put into the edge cases now will save time when we move onto the next step!
As for the tests that clear the reqs/resps maps before the next batch of assertions, it might be helpful to split each of those into their own test case (TEST(MqttStitchTest, XXX) macro). I assume the map .clears() are there to prevent the earlier "test cases" from causing unrelated failures for the later test code. In my opinion, that is a signal that they should be independent cases and any shared setup should be refactored into another function.
src/stirling/source_connectors/socket_tracer/protocols/mqtt/test_utils.h
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher.cc
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher.cc
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher.cc
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher_test.cc
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher_test.cc
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher_test.cc
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher_test.cc
Outdated
Show resolved
Hide resolved
src/stirling/source_connectors/socket_tracer/protocols/mqtt/stitcher_test.cc
Show resolved
Hide resolved
Signed-off-by: Chinmay <[email protected]>
ChinmayaSharma-hue
temporarily deployed
to
pr-actions-approval
GitHub Actions
Inactive
| } else { | ||
| if (type == message_type_t::kRequest) { | ||
| state->send[std::tuple<uint32_t, uint32_t>(result->header_fields["packet_identifier"], | ||
| result->header_fields["qos"])] = 0; | ||
| } else { | ||
| state->recv[std::tuple<uint32_t, uint32_t>(result->header_fields["packet_identifier"], | ||
| result->header_fields["qos"])] = 0; | ||
| } | ||
| } |
There was a problem hiding this comment.
Is it necessary to store a key in the state for PUBLISH'es that don't have any dups?
The ::operator[] method will insert into the map if the key doesn't exist. So I believe the += operator will work as expected without the non dup branch.
| state->send[unique_publish_identifier] > 0) { | ||
| state->send[unique_publish_identifier] -= 1; | ||
| req_frame.consumed = true; | ||
| continue; | ||
| } | ||
|
|
||
| if (req_frame.type == message_type_t::kResponse && | ||
| state->recv[unique_publish_identifier] > 0) { | ||
| state->recv[unique_publish_identifier] -= 1; |
There was a problem hiding this comment.
Related to my earlier comment about avoiding excess inserts into state->{recv,send}. Can we use ::find() instead of ::operator[]?
| {0xc0, 0xd0}, | ||
| // AUTH to AUTH |
There was a problem hiding this comment.
I thought we decided that we were skipping supporting AUTH frames for now. If that changed, I'm ok with keeping this. However, I'd like to add a comment explaining what the gap is (if any) while it's still fresh in our minds.
There was a problem hiding this comment.
I decided to remove the mapping from the map itself, and a TODO before the logic of skipping AUTH matching in the stitcher function.
…ckets Signed-off-by: Chinmay <[email protected]>
ChinmayaSharma-hue
had a problem deploying
to
pr-actions-approval
GitHub Actions
Error
Signed-off-by: Chinmay <[email protected]>
ChinmayaSharma-hue
temporarily deployed
to
pr-actions-approval
GitHub Actions
Inactive
Signed-off-by: Chinmay <[email protected]>
ChinmayaSharma-hue
temporarily deployed
to
pr-actions-approval
GitHub Actions
Inactive
Signed-off-by: Chinmay <[email protected]>
ChinmayaSharma-hue
temporarily deployed
to
pr-actions-approval
GitHub Actions
Inactive
Summary: This PR adds the stitcher component of MQTT (v5), a newly added protocol.
Related issues: #341
Type of change: /kind feature
Test Plan: Added tests