Tuesday, May 24, 2005

AppCrash

AppCrash is a interesting way of collecting information on processes which got SIGBUS or SIGSEGV. It uses DTrace and lets you decide what information is going to be extracted and send to support from a just crashed application. Basically you can do the same with core files - the advantage here is automation. If you are more advanced sys admin then in most cases you don't need AppCrash. If you are not and sending core file is a problem then it's a way to go. Maybe it's a good framework on desktops to provide something similar as in Windows.

2 comments:

Greg Nakhimovsky said...

Robert,

Thanks for sharing your thoughts about our article. Here are my comments. (Note that I speak for myself, not for my employer.)

> Basically you can do the same with core files - the advantage here is automation.

Another advantage is that you can avoid all potential problems associated with core files described in the article.

> If you are more advanced sys admin then in most cases you don't need AppCrash.

I think even the most advanced sys admin can benefit from it.

> Maybe it's a good framework on desktops to provide something similar as in Windows.

Yes, but note that AppCrash can provide a better framework than the one in Windows, due to its more flexible arrangements regarding user control, security, etc.

-Greg

milek said...

I agree. Framework is great - 'coz bascally you can do anything you want in a simple manner. Important thing is to provide scripts which would relay on that framework and would easly integrate on desktop/servers.
Of course that more advanced SAs can benefit from it - especially in large enviroments where such automation would be helpful. I was trying to say that with mission critical systems it all will edn-up with manually analyzing core files, etc. anyway.