1862730 - update stackwalker to rust-minidump v0.20.0

Status: RESOLVED FIXED

(Socorro :: Processor, task, P2)

Description

[Will Kahn-Greene [:willkg] ET needinfo? me]

v0.19.0 just released: https://github.com/rust-minidump/rust-minidump/blob/main/RELEASES.md#version-0190-2023-11-02

We're currently using a version of rust-minidump that's between the v0.18.0 and v0.19.0 releases. We should update to v0.19.0 both so we can be on a release and also to pick up recent things.

This change might warrant some additional work:

Remove support for the ThreadIdNameMapping annotation which is unused since we added support for the thread names stream.

That might be the only reason we pass in the crash annotations as a JSON file. If so, we can stop doing that when we upgrade.

Comment 1

[Will Kahn-Greene [:willkg] ET needinfo? me]

v0.19.1 just released: https://github.com/rust-minidump/rust-minidump/releases/tag/v0.19.1

We should update to that.

Comment 2

[Gabriele Svelto [:gsvelto]]

Amending this since I just cut a new release. This new version contains a few fixes for crashes like this one or this one where the crash reason is not printed correctly (we spit out some hex number where a human-readable value should have been). More importantly it also contains the fix Daniel made for bug 1873529.

Comment 3

[Will Kahn-Greene [:willkg] ET needinfo? me]

Attachment: [mozilla-services/socorro-stackwalk] bug-1862730: update stackwalker to v0.20.0 (#12)

Comment 4

[Will Kahn-Greene [:willkg] ET needinfo? me]

willkg merged PR [mozilla-services/socorro-stackwalk]: bug-1862730: update stackwalker to v0.20.0 (#12) in f588fef.

Comment 5

[Will Kahn-Greene [:willkg] ET needinfo? me]

Attachment: [mozilla-services/socorro] bug-1862730: update stackwalker to v0.20.0 (#6537)

Comment 6

[Will Kahn-Greene [:willkg] ET needinfo? me]

willkg merged PR [mozilla-services/socorro]: bug-1862730: update stackwalker to v0.20.0 (#6537) in 6c357c7.

Comment 7

[Will Kahn-Greene [:willkg] ET needinfo? me]

They just did a rust-minidump v0.21.0 release:

https://github.com/rust-minidump/rust-minidump/releases/tag/v0.21.0

We should pick that up because it includes the change from symbolic to wholesyms which parses symbols files faster. Updating this bug accordingly and passing it to Relud to do the v0.20.0 to v0.21.0 update.

Comment 8

[Daniel Thorn [:relud]]

Attachment: [mozilla-services/socorro-stackwalk] bug-1862730: update stackwalker to v0.21.0 (#13)

Comment 9

[Daniel Thorn [:relud]]

v0.21.0 has a blocking regression (rust-minidump#967):

failure on v0.21.0:

$ build/bin/minidump-stackwalk --evil-json=regr/crashdata/raw_crash/20240228/af300837-1e4e-44b1-b8d0-e0a6b0240228 --symbols-cache=symbols_cache/cache --symbols-tmp=symbols_cache/tmp --no-color --symbols-url=https://symbols.mozilla.org --output-file=regr/20240228_234147/nocache/output.af300837-1e4e-44b1-b8d0-e0a6b0240228.json --json --verbose=error regr/crashdata/upload_file_minidump/af300837-1e4e-44b1-b8d0-e0a6b0240228
ERROR Panic - A panic occurred at minidump-stackwalk/src/main.rs:430: need system info: StreamNotFound

behavior on v0.20.0:

$ build/bin/minidump-stackwalk --evil-json=regr/crashdata/raw_crash/20240228/af300837-1e4e-44b1-b8d0-e0a6b0240228 --symbols-cache=symbols_cache/cache --symbols-tmp=symbols_cache/tmp --no-color --symbols-url=https://symbols.mozilla.org --output-file=regr/20240228_234147/nocache/output.af300837-1e4e-44b1-b8d0-e0a6b0240228.json --json --verbose=error regr/crashdata/upload_file_minidump/af300837-1e4e-44b1-b8d0-e0a6b0240228
ERROR MissingThreadList - Error processing dump: The thread list stream was not found

Comment 10

[Will Kahn-Greene [:willkg] ET needinfo? me]

Given the panic in v0.21.0, I'm going to re-focus this back to v0.20.0 and we can do the v0.21.0 update later in a different bug when the panic is sorted out.

Comment 11

[Will Kahn-Greene [:willkg] ET needinfo? me]

Also, the update to v0.20.0 deployed to stage. I looked through processor_notes values for stackwalker problems and also looked at the Grafana panel showing stackwalker outcomes. There's no obvious change in stackwalker outcomes and exit codes. I think the update looks good.

Comment 12

[Daniel Thorn [:relud]]

(In reply to Will Kahn-Greene [:willkg] ET needinfo? me from comment #10)

Given the panic in v0.21.0, I'm going to re-focus this back to v0.20.0 and we can do the v0.21.0 update later in a different bug when the panic is sorted out.

obsoleted the v0.21.0 attachment and reassigned back to you

Comment 13

[Will Kahn-Greene [:willkg] ET needinfo? me]

This deployed to prod in bug #1882847. Stackwalker exit codes panel in Grafana looks fine. No stackwalker-related issues in Sentry. I think we're good. Marking as FIXED.