image

🎉 🎉 RAPIDMINER 9.10 IS OUT!!! 🎉🎉

Download the latest version helping analytics teams accelerate time-to-value for streaming and IIOT use cases.

CLICK HERE TO DOWNLOAD

How to automatically end processes if they run too long?

kaymankayman Member Posts: 651   Unicorn
edited December 2018 in Product Feedback - Resolved

Hi, since a few days I am experiencing issues with my server where some processed just keep running without finalizing. Where typically they take like 20 minutes or so they now keep runnig indefinitly. As soon as I have one flow giving issues the next one started will also run for forever and as the server only allows to have a few processes running together the rest of my processes just keep waiting screwing up the whole workflow.

 

While I suspect our internal network to be the main cause for the initial process to go idle I was wondering if there are ways to make a process stop automatically if it is running for more than a given time. Like an operator that states 'if I am still running after one hour just kill this process'

 

I never experienced this behaviour before, in the past (7x) I sometimes had processes that did not behave but they were terminated by the server, it seems that this no longer is the case and this is causing us severe problems. Now I noticed it twice in one week (random processes) so I need to find a solution that would allow me to end processes that take too long so they do not impact my workflows too much

0
0 votes

Declined · Last Updated

no comments or votes in over a year - closing this idea for now. Please comment if still relevant.

Comments

  • mschmitzmschmitz Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 3,053  RM Data Scientist

    Hi @kayman,

    a RapidMiner process has always a macro called process_start. It feels like the easiest way would be to check this and throw an exception if this takes to long? But this only works if it is not one operator running too long.

    In this case the logic needs to be in the server.

     

    Best,

    Martin

    - Head of Data Science Services at RapidMiner -
    Dortmund, Germany
  • kaymankayman Member Posts: 651   Unicorn

    Seems like a good starting point indeed, but how would that work in practice ? I could imagine having a check after a given operator to see if I am exceeding my timeframe, but if I reach that point it also means there is probably no real issue (apart from being slow).

     

    The issue I faced is that processes just hang, and seem to do so for infinity unless I stop them directly on the server.

  • mschmitzmschmitz Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 3,053  RM Data Scientist

    Hi,

     

    i've had a few chats internally. I will open up a ticket for a wall time as a setting of a queue.

     

    Best,

    Martin

     

    CC: @sgenzer

    - Head of Data Science Services at RapidMiner -
    Dortmund, Germany
  • sgenzersgenzer 12Administrator, Moderator, Employee, RapidMiner Certified Analyst, Community Manager, Member, University Professor, PM Moderator Posts: 2,959  Community Manager

    yup. Pushing to Product Ideas. @mschmitz can you please ping me the ticket # so I can follow it?

     

    Thanks.


    Scott

     

  • sgenzersgenzer 12Administrator, Moderator, Employee, RapidMiner Certified Analyst, Community Manager, Member, University Professor, PM Moderator Posts: 2,959  Community Manager
Sign In or Register to comment.