In this doc, we introduce the crash reporter in Nebulas, which is used to collect crash reports in Nebulas and send it back to Nebulas Team, so the whole community can help improving the quality of Nebulas.
We, the Nebulas Team and the Nebulas community, always try our best to ensure the stability of Nebulas, since people put their faith and properties on it. That means critical bugs are unacceptable, and we are aware of that. However, we can't blindly think Nebulas is stable enough or there won't be any bugs. Thus, we have plan B, the crash reporter, to collect crash report and send it back to Nebulas community. We hope the whole community can leverage the crash reports and keep improving Nebulas.
Using crash reporter is a very common practice. For example, Microsoft Windows includes a crash reporting service called Windows Error Reporting that prompts users to send crash reports to Microsoft for online analysis. The information goes to a central database run by Microsoft. Apple also involves a standard crash reporter in macOS, named Crash Reporter. The Crash Reporter can send the crash logs to Apple Inc, for their engineers to review. Opensource community also have their own crash reporter, like Bug Buddy for Gnome, Crashpad for Chrome, Talkback for Mozilla, and etc.
In Nebulas, the crash reporter just works like the other crash reporters. It's aware of the crash, collects necessary information about the crash, and sends it back the Nebulas server. The server is hosted by Nebulas, and accessible for the whole community.
As a opensource, decentralized platform, we are aware of that the crash reporter may violate some users' privacy concern. Thus, we remove all private information in the crash report, like the user name, user id, user's home path and IP address. Furthermore, the crash reporter is optional and users may choose close it if users still have some concerns.
To enable or disable the crash reporter, you need to look into the
configuration file, config.conf, and change enable_crash_reporter to true
to enable it, while false to disable it.
In this section, we would like to share some tech details. If you are not interested in the details, you can ignore this section.
The crash reporter is actually a daemon process, which is started by neb.
When starting the crash reporter, neb will tell it the process id (pid) of
neb process, and the crash file path. For the crash reporter, it will
periodically check if the neb process and the crash file exists. At the time
it finds the crash file, it will eliminate the private information and send it
back to Nebulas.
Currently, the crash report is generated by the stderr output from neb.
We'd like the work with the whole community to collect detailed information in
the future.