Instrumenting additional Akka server methods #6243
Conversation
Add support for auto-instrumentation of Akka's IncomingConnection.handleWith and HttpExt.bindAndHandle methods. Both methods utilize a custom BidiGraphStage to wrap a request in instrumentation. A PinnedDispatcher is utilized to make sure the graph stage runs on the same thread for both request and response handling, preventing scope leakage. Co-authored-by: Jeremy Ingle <[email protected]>
* Added Akka HTTP Unmarshaller instrumentation to fix Context loss with 10.2 * Fix tests, all Akka tests are now successful Co-authored-by: Zachary Sistrunk <[email protected]>
We found out that the Unmarshal context loss is actually caused by a larger context loss issue in akka-streams. We'll aim to fix this via a separate future pull request.
|
|
Just confirming that the CLA is still in progress internally. I confirmed with my manager that it's still in the pipeline, it's just taking a while. It's the first time we've contributed to open source, so there's no existing protocol for this kind of request to go through. |
|
hi @zackman0010! any objection to closing this (just trying to clean up the backlog a little), and you can re-open when you are able to sign the CLA? I think if I close it you won't be able to re-open it, but if you close it, I think you will be able to re-open yourself. And of course you can always ping on the closed PR and one of us will re-open it if that fails. also, I'm curious if DCO would be easier for you/your company compared to SLA, check out recent discussion going on about CLA vs DCO in the OpenTelemetry community repo: open-telemetry/community#1252 |
|
@trask Sure, we can do that. I'm sorry again that we haven't had any progress for so long. Our manager has been caught up in another project for several months, and without someone actively working on getting the CLA signed the whole process is stuck. It's a shame because we were very excited to contribute. I'll bring it back up again this week and see if we can get it picked back up, but if that doesn't work I'll go ahead and close the MR. As for CLA vs DCO, I don't know enough about the legal side of things to comment, but I imagine that it would run into the same issues here. After reading the thread, though, my personal opinion as a developer is that I like the suggested hybrid approach. I don't have a problem with signing the CLA, and I'd rather not risk forgetting to sign off on a commit and having to rebase the branch to fix it. |
|
As discussed, closing this PR until our CLA is taken care of. |
Previous PR #6109. Opened a new PR to use a different branch in my fork so that my fork can have a clean
main.Add support for auto-instrumentation of Akka's IncomingConnection.handleWith and HttpExt.bindAndHandle methods. Both methods utilize a custom BidiGraphStage to wrap a request in instrumentation. A PinnedDispatcher is utilized to make sure the graph stage runs on the same thread for both request and response handling, preventing scope leakage.
Fixes #6081 and #5137
Co-authored-by: Jeremy Ingle [email protected]