Do you want to know how the Tessel team identifies and fixes problems in the Tessel command line interface (CLI) without requiring users to reach out to the Tessel team? Say hello to the Tessel 2 crash reporter.
The Tessel 2 crash reporter is a web service which automatically collects anonymized crash reports when a crash is detected in the Tessel 2’s CLI. Given that the CLI is one of the primary ways of interacting with your Tessel, it is important that the CLI be extremely stable and a joy to use.
This project started during the early days of the CLI. We wanted to make sure that the team was proactively looking at and fixing issues as users encountered them. We wanted to make sure that the onboarding experience was as smooth as possible for new users.
How it works
Every time you use the Tessel 2 CLI, a central controller dispatches your command to an appropriate task handler. The crash reporter automatically checks for uncaught
Errors and unresolved
Promises in the
node process that’s executing the task.
Once the crash reporter detects a crash, it removes all personally identifiable information from the stack trace, and uploads it to the Crash reporter web service with the user’s permission. After the user opts-in the first time, subsequent crashes are automatically reported. Below is an example of what that interaction looks like:
ERR! Detected CLI crash [Error: Testing the crash reporter] Error: Testing the crash reporter at CrashReporter.off.CrashReporter.on.CrashReporter.sanitize.CrashReporter.sanitize.redactions.CrashReporter.prompt.CrashReporter.submit.CrashReporter.prompt.then.CrashReporter.post.CrashReporter.status.CrashReporter.test (/Users/rahulrav/Workspaces/Tessel2/CLI/lib/crash-reporter.js:177:25) at process._tickCallback (node.js:406:9) at Function.Module.runMain (module.js:449:11) at startup (node.js:141:18) at node.js:933:3 INFO Detected a crash in the CLI. Submit crash (y/n) ? If yes(y), subsequent crashes will be submitted automatically.
We put a great deal of thought into potential privacy issues, and decided to make this feature opt-in. The user can also turn off crash reporting simply by using one of the below options.
Usage: t2 crash-reporter [options] Options: --off Disable the Crash Reporter. --on Enable the Crash Reporter. --test Test the Crash Reporter. Configure the Crash Reporter.
The Crash reporter web service collects the crash report, and de-duplicates it using the Sim Hash algorithm. Sim Hash looks at the contents of the crash report and computes a unique signature (called the
fingerprint) and also keeps track of the
frequency of the occurence of the report. The service also provides a way to look at
Trending crashes and also provides a full text
Search API. The service itself is built using Google App Engine and the Python Runtime. The full source code for the service is here.
Hi from tessel-crashbot
Recently, we landed a really cool feature in the Crash reporter service. In an effort to simplify the worflow of the Tessel team, which manages all feature requests and bug-fixes using GitHub issues; the Crash reporter service uses GitHub’s API to automatically create and manage issues for every crash report. Anytime there is a new crash in the wild, the ever friendly
tessel-crashbot is ready to help proactively keep track of the problem. The
tessel-crashbot also updates the GitHub issue via comments when it sees an uptick in the frequency of a crash. This helps the team identify the severity of a crash report.