Uploaded image for project: 'Jenkins'
  1. Jenkins
  2. JENKINS-42993

The activity page will hang if too many jobs are run

    • Icon: Bug Bug
    • Resolution: Unresolved
    • Icon: Minor Minor
    • blueocean-plugin

      Problem

      When looking at the activity view when there are lots of jobs running the page will eventually lockup. 

      Steps to Reproduce

      This will setup a job that will run all the time, one instance of the job after the next.

      This is the pipeline script I was using but I am fairly sure that it doesn't matter all that much what the job is doing other than for timing:

      stage('one') {
       echo 'step one'
       }
       stage('two') {
       echo 'step two'
       }
       stage('three') {
       sleep 1
       echo 'step three'
       }
       stage('four') {
       echo 'step four'
       }
      

      Next go to the Blue Ocean activity view and wait for a few minutes. (This goes very fast if you remove the sleep.)

      After those few minutes your browser window should stop responding... And you may see that the browser process is using up a large amount of CPU time.

       

      Notes

      I can split this up into three tickets if you guys want. I am not sure how related they are at this point.

      When I slowed down the job run time with the above sleep the page still locked up which seems to imply a problem with the number of items not the rate that the items are updating. 

      When I slowed down the job I also noticed that the top line item's time kept increasing even though the reported time for the job when it finished was constant. 

      Finally, I noticed that there was a non-zero rate of jobs failing to update in BO. This may be related to the above issue or it could be that the number of jobs run surfaces the issue. Here is a screen cap of the issue in the UI:

          [JENKINS-42993] The activity page will hang if too many jobs are run

          Zachary Joyner created issue -
          James Dumay made changes -
          Description Original: Problem:

          When looking at the activity view when there are lots of jobs running the page will eventually lockup. 

           

          Steps to Reproduce:

          !Screenshot_20170321_154600.png!

          This will setup a job that will run all the time, one instance of the job after the next.

          This is the pipeline script I was using but I am fairly sure that it doesn't matter all that much what the job is doing other than for timing:

          stage('one') \{
           echo 'step one'
          }
          stage('two') \{
           echo 'step two'
          }
          stage('three') \{
           sleep 1
           echo 'step three'
          }
          stage('four') \{
           echo 'step four'
          }

          Next go to the Blue Ocean activity view and wait for a few minutes. (This goes very fast if you remove the sleep.)

          After those few minutes your browser window should stop responding... And you may see that the browser process is using up a large amount of CPU time.

           

          Notes And Variants:

          I can split this up into three tickets if you guys want. I am not sure how related they are at this point.

          When I slowed down the job run time with the above sleep the page still locked up which seems to imply a problem with the number of items not the rate that the items are updating. 

          When I slowed down the job I also noticed that the top line item's time kept increasing even though the reported time for the job when it finished was constant. 

          !Wierd timing bug in BO.png!

          Finally, I noticed that there was a non-zero rate of jobs failing to update in BO. This may be related to the above issue or it could be that the number of jobs run surfaces the issue. Here is a screen cap of the issue in the UI:

          !Failed job in BO.png!
          New: Problem:

          When looking at the activity view when there are lots of jobs running the page will eventually lockup. 

           

          Steps to Reproduce:

          !Screenshot_20170321_154600.png|width=421,height=292!

          This will setup a job that will run all the time, one instance of the job after the next.

          This is the pipeline script I was using but I am fairly sure that it doesn't matter all that much what the job is doing other than for timing:

          stage('one') \{
           echo 'step one'
           }
           stage('two') \{
           echo 'step two'
           }
           stage('three') \{
           sleep 1
           echo 'step three'
           }
           stage('four') \{
           echo 'step four'
           }

          Next go to the Blue Ocean activity view and wait for a few minutes. (This goes very fast if you remove the sleep.)

          After those few minutes your browser window should stop responding... And you may see that the browser process is using up a large amount of CPU time.

           

          Notes And Variants:

          I can split this up into three tickets if you guys want. I am not sure how related they are at this point.

          When I slowed down the job run time with the above sleep the page still locked up which seems to imply a problem with the number of items not the rate that the items are updating. 

          When I slowed down the job I also noticed that the top line item's time kept increasing even though the reported time for the job when it finished was constant. 

          !Wierd timing bug in BO.png!

          Finally, I noticed that there was a non-zero rate of jobs failing to update in BO. This may be related to the above issue or it could be that the number of jobs run surfaces the issue. Here is a screen cap of the issue in the UI:

          !Failed job in BO.png!
          James Dumay made changes -
          Description Original: Problem:

          When looking at the activity view when there are lots of jobs running the page will eventually lockup. 

           

          Steps to Reproduce:

          !Screenshot_20170321_154600.png|width=421,height=292!

          This will setup a job that will run all the time, one instance of the job after the next.

          This is the pipeline script I was using but I am fairly sure that it doesn't matter all that much what the job is doing other than for timing:

          stage('one') \{
           echo 'step one'
           }
           stage('two') \{
           echo 'step two'
           }
           stage('three') \{
           sleep 1
           echo 'step three'
           }
           stage('four') \{
           echo 'step four'
           }

          Next go to the Blue Ocean activity view and wait for a few minutes. (This goes very fast if you remove the sleep.)

          After those few minutes your browser window should stop responding... And you may see that the browser process is using up a large amount of CPU time.

           

          Notes And Variants:

          I can split this up into three tickets if you guys want. I am not sure how related they are at this point.

          When I slowed down the job run time with the above sleep the page still locked up which seems to imply a problem with the number of items not the rate that the items are updating. 

          When I slowed down the job I also noticed that the top line item's time kept increasing even though the reported time for the job when it finished was constant. 

          !Wierd timing bug in BO.png!

          Finally, I noticed that there was a non-zero rate of jobs failing to update in BO. This may be related to the above issue or it could be that the number of jobs run surfaces the issue. Here is a screen cap of the issue in the UI:

          !Failed job in BO.png!
          New: *Problem*

          When looking at the activity view when there are lots of jobs running the page will eventually lockup. 

          *Steps to Reproduce*

          !Screenshot_20170321_154600.png|width=421,height=292!

          This will setup a job that will run all the time, one instance of the job after the next.

          This is the pipeline script I was using but I am fairly sure that it doesn't matter all that much what the job is doing other than for timing:

          {code}
          stage('one') {
           echo 'step one'
           }
           stage('two') {
           echo 'step two'
           }
           stage('three') {
           sleep 1
           echo 'step three'
           }
           stage('four') {
           echo 'step four'
           }
          {code}

          Next go to the Blue Ocean activity view and wait for a few minutes. (This goes very fast if you remove the sleep.)

          After those few minutes your browser window should stop responding... And you may see that the browser process is using up a large amount of CPU time.

           

          Notes And Variants:

          I can split this up into three tickets if you guys want. I am not sure how related they are at this point.

          When I slowed down the job run time with the above sleep the page still locked up which seems to imply a problem with the number of items not the rate that the items are updating. 

          When I slowed down the job I also noticed that the top line item's time kept increasing even though the reported time for the job when it finished was constant. 

          !Wierd timing bug in BO.png!

          Finally, I noticed that there was a non-zero rate of jobs failing to update in BO. This may be related to the above issue or it could be that the number of jobs run surfaces the issue. Here is a screen cap of the issue in the UI:

          !Failed job in BO.png!
          James Dumay made changes -
          Description Original: *Problem*

          When looking at the activity view when there are lots of jobs running the page will eventually lockup. 

          *Steps to Reproduce*

          !Screenshot_20170321_154600.png|width=421,height=292!

          This will setup a job that will run all the time, one instance of the job after the next.

          This is the pipeline script I was using but I am fairly sure that it doesn't matter all that much what the job is doing other than for timing:

          {code}
          stage('one') {
           echo 'step one'
           }
           stage('two') {
           echo 'step two'
           }
           stage('three') {
           sleep 1
           echo 'step three'
           }
           stage('four') {
           echo 'step four'
           }
          {code}

          Next go to the Blue Ocean activity view and wait for a few minutes. (This goes very fast if you remove the sleep.)

          After those few minutes your browser window should stop responding... And you may see that the browser process is using up a large amount of CPU time.

           

          Notes And Variants:

          I can split this up into three tickets if you guys want. I am not sure how related they are at this point.

          When I slowed down the job run time with the above sleep the page still locked up which seems to imply a problem with the number of items not the rate that the items are updating. 

          When I slowed down the job I also noticed that the top line item's time kept increasing even though the reported time for the job when it finished was constant. 

          !Wierd timing bug in BO.png!

          Finally, I noticed that there was a non-zero rate of jobs failing to update in BO. This may be related to the above issue or it could be that the number of jobs run surfaces the issue. Here is a screen cap of the issue in the UI:

          !Failed job in BO.png!
          New: *Problem*

          When looking at the activity view when there are lots of jobs running the page will eventually lockup. 

          *Steps to Reproduce*

          !Screenshot_20170321_154600.png|thumbnail!

          This will setup a job that will run all the time, one instance of the job after the next.

          This is the pipeline script I was using but I am fairly sure that it doesn't matter all that much what the job is doing other than for timing:

          {code}
          stage('one') {
           echo 'step one'
           }
           stage('two') {
           echo 'step two'
           }
           stage('three') {
           sleep 1
           echo 'step three'
           }
           stage('four') {
           echo 'step four'
           }
          {code}

          Next go to the Blue Ocean activity view and wait for a few minutes. (This goes very fast if you remove the sleep.)

          After those few minutes your browser window should stop responding... And you may see that the browser process is using up a large amount of CPU time.

           

          Notes And Variants:

          I can split this up into three tickets if you guys want. I am not sure how related they are at this point.

          When I slowed down the job run time with the above sleep the page still locked up which seems to imply a problem with the number of items not the rate that the items are updating. 

          When I slowed down the job I also noticed that the top line item's time kept increasing even though the reported time for the job when it finished was constant. 

          !Wierd timing bug in BO.png!

          Finally, I noticed that there was a non-zero rate of jobs failing to update in BO. This may be related to the above issue or it could be that the number of jobs run surfaces the issue. Here is a screen cap of the issue in the UI:

          !Failed job in BO.png!
          James Dumay made changes -
          Description Original: *Problem*

          When looking at the activity view when there are lots of jobs running the page will eventually lockup. 

          *Steps to Reproduce*

          !Screenshot_20170321_154600.png|thumbnail!

          This will setup a job that will run all the time, one instance of the job after the next.

          This is the pipeline script I was using but I am fairly sure that it doesn't matter all that much what the job is doing other than for timing:

          {code}
          stage('one') {
           echo 'step one'
           }
           stage('two') {
           echo 'step two'
           }
           stage('three') {
           sleep 1
           echo 'step three'
           }
           stage('four') {
           echo 'step four'
           }
          {code}

          Next go to the Blue Ocean activity view and wait for a few minutes. (This goes very fast if you remove the sleep.)

          After those few minutes your browser window should stop responding... And you may see that the browser process is using up a large amount of CPU time.

           

          Notes And Variants:

          I can split this up into three tickets if you guys want. I am not sure how related they are at this point.

          When I slowed down the job run time with the above sleep the page still locked up which seems to imply a problem with the number of items not the rate that the items are updating. 

          When I slowed down the job I also noticed that the top line item's time kept increasing even though the reported time for the job when it finished was constant. 

          !Wierd timing bug in BO.png!

          Finally, I noticed that there was a non-zero rate of jobs failing to update in BO. This may be related to the above issue or it could be that the number of jobs run surfaces the issue. Here is a screen cap of the issue in the UI:

          !Failed job in BO.png!
          New: *Problem*

          When looking at the activity view when there are lots of jobs running the page will eventually lockup. 

          *Steps to Reproduce*

          !Screenshot_20170321_154600.png|thumbnail!

          This will setup a job that will run all the time, one instance of the job after the next.

          This is the pipeline script I was using but I am fairly sure that it doesn't matter all that much what the job is doing other than for timing:

          {code}
          stage('one') {
           echo 'step one'
           }
           stage('two') {
           echo 'step two'
           }
           stage('three') {
           sleep 1
           echo 'step three'
           }
           stage('four') {
           echo 'step four'
           }
          {code}

          Next go to the Blue Ocean activity view and wait for a few minutes. (This goes very fast if you remove the sleep.)

          After those few minutes your browser window should stop responding... And you may see that the browser process is using up a large amount of CPU time.

           

          Notes And Variants:

          I can split this up into three tickets if you guys want. I am not sure how related they are at this point.

          When I slowed down the job run time with the above sleep the page still locked up which seems to imply a problem with the number of items not the rate that the items are updating. 

          When I slowed down the job I also noticed that the top line item's time kept increasing even though the reported time for the job when it finished was constant. 

          !Wierd timing bug in BO.png|thumbnail!

          Finally, I noticed that there was a non-zero rate of jobs failing to update in BO. This may be related to the above issue or it could be that the number of jobs run surfaces the issue. Here is a screen cap of the issue in the UI:

          !Failed job in BO.png|thumbnail!
          James Dumay made changes -
          Description Original: *Problem*

          When looking at the activity view when there are lots of jobs running the page will eventually lockup. 

          *Steps to Reproduce*

          !Screenshot_20170321_154600.png|thumbnail!

          This will setup a job that will run all the time, one instance of the job after the next.

          This is the pipeline script I was using but I am fairly sure that it doesn't matter all that much what the job is doing other than for timing:

          {code}
          stage('one') {
           echo 'step one'
           }
           stage('two') {
           echo 'step two'
           }
           stage('three') {
           sleep 1
           echo 'step three'
           }
           stage('four') {
           echo 'step four'
           }
          {code}

          Next go to the Blue Ocean activity view and wait for a few minutes. (This goes very fast if you remove the sleep.)

          After those few minutes your browser window should stop responding... And you may see that the browser process is using up a large amount of CPU time.

           

          Notes And Variants:

          I can split this up into three tickets if you guys want. I am not sure how related they are at this point.

          When I slowed down the job run time with the above sleep the page still locked up which seems to imply a problem with the number of items not the rate that the items are updating. 

          When I slowed down the job I also noticed that the top line item's time kept increasing even though the reported time for the job when it finished was constant. 

          !Wierd timing bug in BO.png|thumbnail!

          Finally, I noticed that there was a non-zero rate of jobs failing to update in BO. This may be related to the above issue or it could be that the number of jobs run surfaces the issue. Here is a screen cap of the issue in the UI:

          !Failed job in BO.png|thumbnail!
          New: *Problem*

          When looking at the activity view when there are lots of jobs running the page will eventually lockup. 

          *Steps to Reproduce*

          !Screenshot_20170321_154600.png|thumbnail!

          This will setup a job that will run all the time, one instance of the job after the next.

          This is the pipeline script I was using but I am fairly sure that it doesn't matter all that much what the job is doing other than for timing:

          {code}
          stage('one') {
           echo 'step one'
           }
           stage('two') {
           echo 'step two'
           }
           stage('three') {
           sleep 1
           echo 'step three'
           }
           stage('four') {
           echo 'step four'
           }
          {code}

          Next go to the Blue Ocean activity view and wait for a few minutes. (This goes very fast if you remove the sleep.)

          After those few minutes your browser window should stop responding... And you may see that the browser process is using up a large amount of CPU time.

           

          *Notes*

          I can split this up into three tickets if you guys want. I am not sure how related they are at this point.

          When I slowed down the job run time with the above sleep the page still locked up which seems to imply a problem with the number of items not the rate that the items are updating. 

          When I slowed down the job I also noticed that the top line item's time kept increasing even though the reported time for the job when it finished was constant. 

          !Wierd timing bug in BO.png|thumbnail!

          Finally, I noticed that there was a non-zero rate of jobs failing to update in BO. This may be related to the above issue or it could be that the number of jobs run surfaces the issue. Here is a screen cap of the issue in the UI:

          !Failed job in BO.png|thumbnail!
          James Dumay made changes -
          Priority Original: Major [ 3 ] New: Minor [ 4 ]
          James Dumay made changes -
          Epic Link New: JENKINS-35742 [ 171802 ]
          James Dumay made changes -
          Labels New: technical-debt
          James Dumay made changes -
          Epic Link Original: JENKINS-35742 [ 171802 ] New: JENKINS-43952 [ 181484 ]
          James Dumay made changes -
          Sprint New: Blue Ocean 1.3 [ 311 ]
          James Dumay made changes -
          Rank New: Ranked lower

            Unassigned Unassigned
            zjoyner Zachary Joyner
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

              Created:
              Updated: