Fix memory leak reading from python ByteBuffer. #6711
Conversation
|
Thanks for your pull request. The automated tests will run as soon as one of the admins verifies this change is ok to test. |
|
@alamaison |
|
@zj8487 I believe so. I only plotted the server memory usage, but as I was doing so, I saw the client memory usage in htop climb by a similiar amount |
|
@alamaison |
|
This is okay to test. |
|
Looks great to me; @soltanmm will also review. Thanks! |
|
@nathanielmanistaatgoogle |
|
The failures in 'Basic tests' look like flakes:
|
|
@alamaison Yow - thanks for the catch! Looks good to me too. Will mark ready to merge on interop test success. @zj8487 I think we're planning on releasing a 0.14.2 soon. @jtattermusch? |
|
@soltanmm Are the interop tests stuck? |
|
@alamaison Jenkins is being a silly one... |
|
@jtattermusch Got any tips on how to make Jenkins re-run the interop tests on this PR? |
|
@jtattermusch @soltanmm bump |
|
@alamaison I hate to ask you to do this, but, would you mind re-opening the pull request? Jenkins is being inordinately stubborn in spite of being kicked a few times. Else, we can just PR off of your branch again. |
Fixes #5913. `grpc_byte_buffer_reader_next` is documented as 'Caller is responsible for calling gpr_slice_unref on the result', but that wasn't happening. This commit adds the missing call to `gpr_slice_unref`.
|
@soltanmm No problem. I've pushed again. Jenkins is running. |
|
@soltanmm A different set of tests failed this time. Timeouts in the C/C++ tests so definitely not caused by this PR. Are you able to merge as-is or do you want me to push again? |
|
lgtm; marking |
|
Thanks @alamaison! |
Fixes #5913.
grpc_byte_buffer_reader_nextis documented as 'Caller is responsiblefor calling gpr_slice_unref on the result', but that wasn't happening.