1503316 - spidermonkey shell jobs are generating unwanted minidumps

Status: NEW

(Core :: JavaScript Engine, enhancement, P3)

Description

[Steve Fink [:sfink] [:s:]]

Bug 1410528 supposedly set things up so that we would only get minidumps from unexpected crashes, but it seems like something in that mechanism is incorrect -- when a spidermonkey shell job fails, it will have accumulated a few unwanted minidumps (as in, minidumps from tests that were expected to crash). This clutters up the output, and suggests problems that are not problems. See eg bug 1499681.

Comment 1

[Steve Fink [:sfink] [:s:]]

Actually, I was wrong. Bug 1499681 is a desired stack dump -- we *want* stack dumps for timeouts, to help with diagnosing why it's taking so long.

I know I have seen cases where one crash was accompanied with several spurious stack dumps, though, so I believe this bug is still valid. In fact, it's happened within the last week, so I don't think this has been fixed yet.

Comment 2

[Nicolas B. Pierron [:nbp]]

Sounds like a resources saving bug, marking as P3 for now, feel free to promote to P2 / P1 if this is more urgent to fix.

Comment 3

[Steve Fink [:sfink] [:s:]]

(In reply to Nicolas B. Pierron [:nbp] from comment #2)
> Sounds like a resources saving bug, marking as P3 for now, feel free to
> promote to P2 / P1 if this is more urgent to fix.

No, the resources are pretty much irrelevant. The problem is that when you get one real failure, it's confusing to find the correct stack because a bunch of extra stacks are emitted into the logs.

But I'm fine with P3.