Standalone spark cluster dead nodes

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Standalone spark cluster dead nodes

Debasish Das
Hi,

I have been running standalone spark cluster but sometimes I do see dead nodes. The physical machines are not dead but the JVM worker dies.

Is there a methodology which automatically restart worker JVM if it dies ?

Thanks.
Deb

Reply | Threaded
Open this post in threaded view
|

Re: Standalone spark cluster dead nodes

Andrew Ash
Do you have stacktraces or other errors for the workers' deaths?

This is a great case for using the Tanuki service wrapper that can be configured to automatically restart JVMs that die.

Andrew


On Thu, Jan 2, 2014 at 4:38 PM, Debasish Das <[hidden email]> wrote:
Hi,

I have been running standalone spark cluster but sometimes I do see dead nodes. The physical machines are not dead but the JVM worker dies.

Is there a methodology which automatically restart worker JVM if it dies ?

Thanks.
Deb