JPPF Issue Tracker
star_faded.png
Please log in to bookmark issues
feature_request_small.png
OPEN  Feature request JPPF-462  -  Node temperature
Posted Jul 15, 2016 - updated Jul 26, 2016
action_vote_minus_faded.png
0
Votes
action_vote_plus_faded.png
lolo4j (lolocohen) has been working on this issue since July 26, 2016 (01:31)
Issue details
  • Type of issue
    Feature request
  • Status
     
    New
  • Assigned to
     lolo4j
  • Type of bug
    Not triaged
  • Likelihood
    Not triaged
  • Effect
    Not triaged
  • Posted by
     beiri22
  • Owned by
    Not owned by anyone
  • Category
    Management / Monitoring
  • Resolution
    Not determined
  • Priority
    Normal
  • Targetted for
    icon_milestones.png JPPF 6.0
Issue description
Hi there, I am using JPPF on a few nodes that are in a closed room. On workload the temperature rises notably. From time to time I check both room and pc temperature manually. But as each computer has some temperature sensors (board, cpu, ...), I would prefer a solution that displays the temperature information in the management console. Each node could read out these information and send it to the driver. What do you think?

#4
Comment posted by
 beiri22
Jul 20, 15:34
The temperature monitoring could be done very easily with a broadcast job. Unfortunately, broadcast jobs are not able to hand back a result to the client. Can you allow broadcast jobs to return a result?
#5
Comment posted by
 lolo4j
icon_reply.pngJul 26, 01:31, in reply to comment #4
You want to be able to monitor the node while tasks are executing in it. If a job is already executing on the node, then you can't run another job at the same time, whether it's a broadcast job or not. The current way of monitoring, using JMX, is what should be used.

In any case, that's not the main problem. The problem with thermal sensors is that there is no standard protocol to access them. It depends on the hardware you have, how the device drivers interact with it, and whether the OS can access the functionality. Furthermore, you'd need some way to access thermal readings from Java, which adds yet another layer of complexity. Plus, hardware specifications change all the time and there would be a huge maintenance cost. Due to all this, I have no intention to implement this specific type of monitoring.

However, what I can propose, and which I think would benefit the whole JPPF community, is an extension point to inject custom monitoring data (i.e. additional columns) in the "JVM health" view of the console. Note that you already have the possiblity to customize the admin console, along with the servers and nodes, to add custom monitoring:

beiri22 wrote:
The temperature monitoring could be done very easily with a
broadcast job. Unfortunately, broadcast jobs are not able to
hand back a result to the client. Can you allow broadcast jobs
to return a result?