Skip to main content

Tools for working with Crash Stats (https://crash-stats.mozilla.org/)

Project description

Command line tools and library for interacting with Crash Stats (https://crash-stats.mozilla.org/).

Code:

https://github.com/willkg/crashstats-tools

Documentation:

Check the README.rst file

Changelog:

Check the HISTORY.rst file

Issue tracker:

https://github.com/willkg/crashstats-tools/issues

License:

MPLv2

Chat:

#crashreporting matrix channel

Community Participation Guidelines:

https://github.com/willkg/crashstats-tools/blob/main/CODE_OF_CONDUCT.md

Installing

crashstats-tools is available on PyPI. You can install it with:

$ pip install --user crashstats-tools

Since it has commands that you want to use, you probably want to install it with pipx:

$ pipx install crashstats-tools

Support

This is a project that I spent some time on because it seemed helpful to me and others. I use it regularly for Crash Stats investigations and reprocessing. However, I’m not you! If you have any problems, please write up an issue in the issue tracker and I’ll get to it when I’m able.

If you use this, like it, appreciate it, or have any positive feeling, please give it a GitHub star. That helps me:

  1. have a proxy for knowing whether it’s being used

  2. prioritize my time working on this project

  3. have a hand-wavey list of users in case I have questions and need to ask for advice

Tools

supersearch

Perform Super Search queries on Crash Stats.

Fetch 10 crash ids for Firefox:

$ supersearch --num=10 --product=Firefox

Fetch all crash reports that have libc2.30.so/E22A1E7AEF7C58504AF2C60A5AD3A7AE0 in the modules_in_stack field:

$ supersearch --modules_in_stack=libc2.30.so/E22A1E7AEF6C58504AF2C60A5AD3A7AE0

This is helpful when you need to reprocess crash reports after uploading symbols for a module that we didn’t have symbols for.

Fetch all crash reports that have libfenix.so in the modules_in_stack field:

$ supersearch --verbose --modules_in_stack='^libfenix.so'

Fetch 57 crash ids that match a Super Search query:

$ supersearch --num=57 --supersearch-url='https://crash-stats.mozilla.org/search/?release_channel=nightly&version=70.0a1&product=Firefox&_sort=-date'

Use single quotes for values so that your shell doesn’t do any shell expansion.

Fetch uuid, product, version, and build_id for crash reports that have “OOM” in the signature:

$ supersearch --_columns=uuid --_columns=product --_columns=version \
    --_columns=build_id --signature='~OOM'

Results are formatted as tab-delimited by default. JSON output is also available.

Note that this doesn’t support Super Search aggregations–use supersearchfacet for that.

See Super Search API documentation for details on notation and fields:

supersearchfacet

Performs facets and daily-facets.

See the breakdown of crash reports by product for the last 7 days:

$ supersearchfacet --_facets=product

See crashes broken down by product and down by day for the last 7 days:

$ supersearchfacet --daily --relative-range=7d --_facets=product

See just Firefox crashes broken down by day for the last 14 days:

$ supersearchfacet --daily --relative-range=14d --_facets=product --product=Firefox

Results are formatted as tab-delimited by default. Markdown and JSON output are also available.

Get the table in Markdown for easy cut-and-paste into Markdown things:

$ supersearchfacet --daily --format=markdown --relative-range=14d --_facets=product \
    --product=Firefox

See Super Search API documentation for details on notation and fields:

fetch-data

Fetch data for specified crash reports.

This lets you download raw crash, dumps, and processed crash from Crash Stats.

Fetch processed crash data for specific crash id:

$ fetch-data --no-raw --no-dumps --processed 723cacd6-1684-420e-a1c7-f04240190731

Fetch raw crash data using supersearch command to generate crash ids:

$ supersearch --product=Firefox --num=10 | \
    fetch-data --raw --no-dumps --no-processed crashdir

reprocess

Let’s you specify crash reports for reprocessing.

Reprocess an individual crash report:

$ reprocess 723cacd6-1684-420e-a1c7-f04240190731

Reprocess crash reports based on a supersearch:

$ supersearch --num=5 | reprocess

API token

For supersearch and fetch-data, you need to use a API token to:

  • download data containing personally identifiable information

  • download security sensitive data

  • get out from the shadow of extreme API use rate limiting

You need an API token for reprocess–it doesn’t work without one.

If you have access, you can generate an API token here:

https://crash-stats.mozilla.org/api/tokens/

Once you have acquired one, set the CRASHSTATS_API_TOKEN environment variable when using crashstats-tools commands.

Remember to abide by the data access policy when using data from Crash Stats! The policy is specified here:

https://crash-stats.mozilla.org/documentation/memory_dump_access/

Use cases

These tools are helpful when downloading data for analysis as well as downloading data to test other tools with.

Example 1

I want to collect a bunch of crash report data to look at possible values of an annotation in Firefox crash reports that’s not available in Super Search, yet.

Since I’m looking just at annotations, all I need is the raw crash.

I would do something like this:

$ mkdir crashdata
$ supersearch --product=Firefox --num=1000 | \
    fetch-data --raw --no-dumps --no-processed crashdata

Then I can use jq or whatever to look at the crash report data in crashdata/raw_crash/.

Example 2

I want to test out a new JIT analysis tool that works on minidump files.

I would write a script like this:

#!/bin/bash

CRASHSTATS_API_TOKEN=foo
DATADIR=./crashdata
CRASHIDS=$(supersearch --product=Firefox --num=1000)

mkdir -p "${DATADIR}"

for crashid in ${CRASHIDS}
do
    echo "crashid ${crashid}"
    fetch-data --raw --dumps --no-processed "${DATADIR}" "${crashid}"

    # Not all crash reports have dumps--we only want to run analysis
    # on the ones that do.
    if [[ -e "crashdata/dump/${crashid}" ]]
    then
        echo "analyze dump ${crashid}..."
        # run my tool on the dump
    fi
done

Example 3

I want to get a list of crash ids for today (2019-07-30) where DOMFissionEnabled exists in the crash report.

I would do this:

$ supersearch --date=">=2019-07-30" --date='<2019-07-31' \
    --dom_fission_enabled='!__null__'

Example 4

I want to see number of crash reports for the last 14 days broken down by day and by product where DOMFissionEnabled exists in the crash report.

I would do this:

$ supersearchfacet --period=daily --format=markdown --relative-range=14d \
    --dom_fission_enabled='!__null__' --_facets=product

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

crashstats-tools-1.4.0.tar.gz (26.5 kB view hashes)

Uploaded Source

Built Distribution

crashstats_tools-1.4.0-py3-none-any.whl (26.8 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page