[
https://issues.apache.org/jira/browse/GEARPUMP-146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Karol Brejna updated GEARPUMP-146:
----------------------------------
Description:
The plan for the experiment is to have a master and one (or more) workers on
"normal" machine (a laptop, a pc) and one worker on the device.
To run this cluster we need:
1) gear.conf for the edge device - with trimmed requirements (memory, number of
slots, maybe metrics disabled) so the worker fits in
2) networking configuration so the workers see the master
(Optionally I could do some ".zip fitness" to have smaller binaries for the
device; something like zip --delete gearpump-2.11-0.8.0.zip
gearpump-2.11-0.8.0/dashboard/* gearpump-2.11-0.8.0/examples/*
gearpump-2.11-0.8.0/lib/services/* gearpump-2.11-0.8.0/lib/storm/*
gearpump-2.11-0.8.0/lib/yarn/*)
Next step is to deploy an example DAG and demonstrate (show in the dashboard)
that some operations are conducted on the raspi worker.
was:
The plan for the experiment is to have a master and one (or more) workers on
"normal" machine (a laptop, a pc) and one worker on the device.
To run this cluster we need:
1) gear.conf for the edge device - with trimmed requirements (memory, number of
slots, maybe metrics disabled) so the worker fits in
2) networking configuration so the workers see the master
(Optionally I oculd do some ".zip fitness" to have smaller binaries for the
device; something like zip --delete gearpump-2.11-0.8.0.zip
gearpump-2.11-0.8.0/dashboard/* gearpump-2.11-0.8.0/examples/*
gearpump-2.11-0.8.0/lib/services/* gearpump-2.11-0.8.0/lib/storm/*
gearpump-2.11-0.8.0/lib/yarn/*)
Next step is to deploy an example DAG and demonstrate (show in the dashboard)
that some operations are conducted on the raspi worker.
> Gearpump on Raspi - "experiment" plan
> -------------------------------------
>
> Key: GEARPUMP-146
> URL: https://issues.apache.org/jira/browse/GEARPUMP-146
> Project: Apache Gearpump
> Issue Type: Task
> Reporter: Karol Brejna
> Priority: Minor
>
> The plan for the experiment is to have a master and one (or more) workers on
> "normal" machine (a laptop, a pc) and one worker on the device.
> To run this cluster we need:
> 1) gear.conf for the edge device - with trimmed requirements (memory, number
> of slots, maybe metrics disabled) so the worker fits in
> 2) networking configuration so the workers see the master
> (Optionally I could do some ".zip fitness" to have smaller binaries for the
> device; something like zip --delete gearpump-2.11-0.8.0.zip
> gearpump-2.11-0.8.0/dashboard/* gearpump-2.11-0.8.0/examples/*
> gearpump-2.11-0.8.0/lib/services/* gearpump-2.11-0.8.0/lib/storm/*
> gearpump-2.11-0.8.0/lib/yarn/*)
> Next step is to deploy an example DAG and demonstrate (show in the dashboard)
> that some operations are conducted on the raspi worker.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
