Author Topic: sending email fails..  (Read 2675 times)

mpursley

  • Full Member
  • ***
  • Posts: 15
sending email fails..
« on: December 11, 2009, 02:01:25 AM »

Hey All,

I have setup my qube sup to send emails with jobs have finished... and I've noticed that sometimes the emails are sent, and other times they are not...

I found that this qbsub command would send the email upon completion...
----------------------------------------
[mp@vfx06] /Applications
$ /Applications/pfx/qube/bin/qbsub  --groups maya --cluster /farm --priority 9000 --cpus 5  --name "render2009: render_test_000007_TEMP.mb" --range 1-5  --mail --address "mpursley@testadd.com" "echo /opt/local/bin/Render2009  -s QB_FRAME_START -e QB_FRAME_START                               /Volumes/VFX/Projects/Treasure_Buddies/Assets/Development/Artist/matt_pursley/sand"
10934
-------------------------------------------


... and this command will not send out the email opon completion...
--------------------------------------------
[mp@vfx06] /Applications
$ /Applications/pfx/qube/bin/qbsub  --groups maya --cluster /farm --priority 9000 --cpus 5  --name "render2009: render_test_000007_TEMP.mb" --range 1-5  --mail --address "mpursley@testadd.com" "echo /opt/local/bin/Render2009  -s QB_FRAME_START -e QB_FRAME_START                               /Volumes/VFX/Projects/Treasure_Buddies/Assets/Development/Artist/matt_pursley/sand1"
10935
---------------------------------------




I will include a .txt file with the .xja files.  This is a reproduceble error that happens the same every time.   And everything is the same between the 2 commands, except for the "1" character at the end.


Anyone have any idea why this would happen?  I have some jobs that can send emails... and other that can't...


Thanks,
Matt




[attachment deleted by admin]

mpursley

  • Full Member
  • ***
  • Posts: 15
Re: sending email fails..
« Reply #1 on: December 11, 2009, 02:03:47 AM »
Here's a screen shot of my shell running the 2 commands back to back... and in the email client I am getting only every other email...






Hey All,

I have setup my qube sup to send emails with jobs have finished... and I've noticed that sometimes the emails are sent, and other times they are not...

I found that this qbsub command would send the email upon completion...
----------------------------------------
[mp@vfx06] /Applications
$ /Applications/pfx/qube/bin/qbsub  --groups maya --cluster /farm --priority 9000 --cpus 5  --name "render2009: render_test_000007_TEMP.mb" --range 1-5  --mail --address "mpursley@testadd.com" "echo /opt/local/bin/Render2009  -s QB_FRAME_START -e QB_FRAME_START                               /Volumes/VFX/Projects/Treasure_Buddies/Assets/Development/Artist/matt_pursley/sand"
10934
-------------------------------------------


... and this command will not send out the email opon completion...
--------------------------------------------
[mp@vfx06] /Applications
$ /Applications/pfx/qube/bin/qbsub  --groups maya --cluster /farm --priority 9000 --cpus 5  --name "render2009: render_test_000007_TEMP.mb" --range 1-5  --mail --address "mpursley@testadd.com" "echo /opt/local/bin/Render2009  -s QB_FRAME_START -e QB_FRAME_START                               /Volumes/VFX/Projects/Treasure_Buddies/Assets/Development/Artist/matt_pursley/sand1"
10935
---------------------------------------




I will include a .txt file with the .xja files.  This is a reproduceble error that happens the same every time.   And everything is the same between the 2 commands, except for the "1" character at the end.


Anyone have any idea why this would happen?  I have some jobs that can send emails... and other that can't...


Thanks,
Matt




[attachment deleted by admin]
« Last Edit: December 11, 2009, 06:07:39 PM by mpursley »

jburk

  • Administrator
  • *****
  • Posts: 493
Re: sending email fails..
« Reply #2 on: December 11, 2009, 08:06:18 PM »
I'm unable to replicate this on my end.

Code: [Select]
jburk-mbPro:~ jburk$ qbsub -req host.os=linux -mail -name "mail test" ls /home/jburk/test
2694
jburk-mbPro:~ jburk$ qbsub -req host.os=linux -mail -name "mail test" ls /home/jburk/test/1
2695
jburk-mbPro:~ jburk$ qbsub -req host.os=linux -mail -name "mail test" ls /home/jburk/test1
2696

I received mail for all 3 jobs.

Check the callbacks for the jobs that fail to send mail.  Look for the one that has a language set to "mail".  It should look something like this:

count       : 1
code        :
language    : mail
triggers    : done-job-self
pid         : 2696
maximum     : -1
user        : jburk
ready       : done-job-self 1
id          : 1


You're interested in the value at the end of the 'ready' line: 1 means the condition to fire the trigger has been met, 0 means not.  If ready == 1, check the 'count', which is how many times the callback actually ran.  It's usually 1, but if it's 0, then it never ran, despite the trigger condition being met.

If the callback is running, you'll need to dig deeper.

Can you build a test job that runs very fast that fails to send mail?  If so, is your supervisor an OSX or linux box, and if you have access to it, you may want to tail the supelog and see what's happening when your job runs. 

Look for lines like:

[Dec 11, 2009 9:04:57] build01 : executing mail callback (2696)2696*1 trigger: done-job-self event: done-job-2696
[Dec 11, 2009 9:04:57] build01 : executing mail callback (2696): done-job-self event: done-job-2696 - begin
[Dec 11, 2009 9:04:57] build01 : sending notification mail to: jburk
[Dec 11, 2009 9:04:59] build01 : sent mail to user: jburk via: mail.pipelinefx.com
[Dec 11, 2009 9:04:59] build01 : executing mail callback (2696): done-job-self event: done-job-2696 - end


If you see the 'sent mail' line, but you never receive the mail, it may be getting caught in a spam trap, either on your end or somewhere between the supervisor and your mailbox.