Investigate flaky test parallel/test-zlib-convenience-methods

Platform: Ubuntu 14.04 (so far)
Subsystem: zlib

https://ci.nodejs.org/job/node-test-commit-linux/21225/nodes=ubuntu1404-32/console:

13:26:19   ---
13:26:19   duration_ms: 0.227
13:26:19   severity: crashed
13:26:19   exitcode: -6
13:26:19   stack: |-
13:26:19     *** Error in `out/Release/node': free(): invalid pointer: 0x0ae0b5f8 ***

Might be helpful to have a core dump/binary pair (@nodejs/build)?

stress: ~~https://ci.nodejs.org/job/node-stress-single-test/1998/nodes=ubuntu1404-64/~~
https://ci.nodejs.org/job/node-stress-single-test/1999/nodes=ubuntu1404-64

Stress test is currently at 5k/10k okay, 0 not okay. What happens if it comes back green? Do we close the issue and accept that this test fails every once in a blue moon?

What happens if it comes back green?

I usually try some other parameters. If it's a parallel test, I add some of its neighbors to the mix (for example parallel/test-z*). An other option is to run it on the same host with the same exact code.

gist of log so we don't lose it https://gist.github.com/refack/0ec2aa4c53cb1e4c89a8f0ce9675f823

Stress test did come back green, so, trying again: https://ci.nodejs.org/job/node-stress-single-test/2000/

Now seeing on that platform (https://ci.nodejs.org/job/node-test-commit-linux/21266/nodes=ubuntu1404-32/console):

11:48:27 *** Error in `python': corrupted size vs. prev_size: 0x085f4498 ***

@nodejs/build I’d say there’s either something wrong with the OS, or the machine might have a genuine hardware issue.

It crashed here as well: https://ci.nodejs.org/job/node-test-commit-linux/21645/nodes=ubuntu1404-32/console

13:10:10 not ok 2046 parallel/test-zlib-convenience-methods
13:10:10   ---
13:10:10   duration_ms: 0.228
13:10:10   severity: crashed
13:10:10   exitcode: -6
13:10:10   stack: |-
13:10:10     *** Error in `out/Release/node': double free or corruption (!prev): 0x0bbf36f0 ***
13:10:10   ...

Yeah. It’s super weird how on the Node.js side, it’s only this one test that’s failing this frequently, when other programs seem to be having memory issues as well…

This hasn't been seen for a while, has it? Should it be closed?

Navigation Menu

Repository navigation

Investigate flaky test parallel/test-zlib-convenience-methods #22628

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Investigate flaky test parallel/test-zlib-convenience-methods #22628

Description

Activity

refack commented on Aug 31, 2018

addaleax commented on Aug 31, 2018

refack commented on Aug 31, 2018

addaleax commented on Sep 1, 2018

addaleax commented on Sep 2, 2018

tniessen commented on Sep 18, 2018

addaleax commented on Sep 18, 2018

Trott commented on Nov 21, 2018

Trott commented on Nov 21, 2018

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions