Jobs crash and become unusable for unclear reasons

Comments

10 comments

  • Avatar
    Rocco Pigneri

    Dear Robert,

    Thank you for providing us with such thorough information on the issues you experienced.  I have taken the liberty of converting your attachment into an in-page post so others can Google for the same issue in the future.

    Would you be able to provide us with your log files?  You can find directions on how to capture those log files here.  Once captured, you can either attach them here, or e-mail them to help@lavastorm.com.

    In the meantime, I will write up some steps on how to resolve your first error.  Please know that we have fixed a lot of similar issues in 3.1.3 (to be released any day now) so an upgrade may help out here.

    Hope that helps,

    Rocco

    0
    Comment actions Permalink
  • Avatar
    Robert Eddy

    Hi Rocco, 

    We're using the Dataverse client on Dataverse Server, what is the directory path for the logs for Server on Linux?  I don't have access to the machine so will have to pass along instructions to the admin.

    Are Dataverse releases (in this case 3.1.3) for Server available at the same time as the Dataverse Desktop?  How will we know when the update is available when using Server?

    Thanks, Rob

    0
    Comment actions Permalink
  • Avatar
    Rocco Pigneri

    Dear Rob,

    The directory for the server logs on Linux are configurable during installation.  However, it defaults to <Dataverse Installation Directory>/site-<DV user port>.  In a default installation of Dataverse 3.1.x, that port number is 7731.

    Both server and desktop installers are released at the same time for all platforms.  Just so you know, DV 3.1.3 was released yesterday so it is already available to you.

    As for announcements of new versions, you can most easily find those on Twitter and LinkedIn.  Our Twitter handle is @YourDataverse.

    I am still working on your original problem and hope to provide a solution soon.

    Hope that helps,

    Rocco

    0
    Comment actions Permalink
  • Avatar
    Rocco Pigneri

    Dear Rob,

    I have looked into this, and I do have a solution.  However, it is very involved.

    Given that, we do believe that upgrading to 3.1.3 may resolve this ;issue for you in a simpler fashion.  Would that be interesting?

    Thank you,

    Rocco

    0
    Comment actions Permalink
  • Avatar
    Robert Eddy

    Hi Rocco, 

    After some issues with the upgrade (it was more of a reinstall since everything was overwritten) and a lot of help (again) from Paul Rasmussen, we successfully upgraded to 3.1.3.

    I'm running a query that should return about 2 million rows, it doesn't come back.  If I click on the node after a while the job crashes. I changed the query to only return 1k rows and it works.  When I remove the restrictive where clause and try again for the 2 million it crashes.

    Sorry, something has gone wrong

    An unexpected problem has caused the application to stop working properly.

    dataref not found: " file:/Data/dataverse/3.1/site-7731/tmp/rob/paceAccountExtract/60451d46-22ef-4388-b273-3cbd45fcd0a4/temp.r00jprn0c.bnymellon.net.27604.0.1498528846531.a3b166b16b2f5792fd6420d9ba1d5db5.brd

     

    0
    Comment actions Permalink
  • Avatar
    Rocco Pigneri

    Dear Robert,

    I'm sorry to hear that you had upgrade troubles, but I'm happy that Paul got you through them.

    As you know, we have put a number of performance increases into our database nodes so you should no longer see this issue in 3.1.4 (due out in a few weeks).

    As for the crash, I haven't been able to reproduce it in 3.1.3 with either a database node or other nodes.  Could you confirm that that issue was in 3.1.3?  Could you also provide me with some more details about how the node crashes?  Are you hitting the stop button, or does the node die on its own?

    Thank you,

    Rocco

    0
    Comment actions Permalink
  • Avatar
    Rocco Pigneri

    Dear Rob,

    Returning to your original post, I just heard back from Engineering about the second issue you reported, the message containing "nodeProperties.stateTypes.PENDING".  This was a display issue and does not affect the outcome of your runs.  It has also been fixed in 3.1.4.

    Hope that helps,

    Rocco

    0
    Comment actions Permalink
  • Avatar
    Robert Eddy

    Hi Rocco, sorry for the delay.  Regarding the crash, yes, I believe I was hitting the stop button.  The jdbc node wasn't returning data so after about 20 or 30 minutes I killed it.  Then changed queries and reran, and probably killed those too.  

    The good news is I no longer have any database issue.  Paul helped me tweak the settings on our Server install and now a 56 minute query comes back in 1 minute!  

    I haven't needed to kill a db query and have not seen the crash since.

    Thanks for all of your help, I think our team is in a good place now.  

    Thanks,

    Rob

    0
    Comment actions Permalink
  • Avatar
    Rocco Pigneri

    Sounds great, Rob!  Glad to hear that it's working!

    Feel free to write back if you have any other questions.

    Rocco

    0
    Comment actions Permalink
  • Avatar
    Robert Eddy

    Ok, spoke to soon.  Another crash for unclear reasons.  This time I was trying to work with the scripting language in a transform node and for whatever reason, the job blew up and won't let me open it.  I've exported and reimported (which I had to do earlier today for unclear reasons) but it hasn't worked.  

    Here's the error message:

    0
    Comment actions Permalink

Please sign in to leave a comment.



Powered by Zendesk