3.99.37 beta problem with process kill after watchdog restart

Started by rapidset, October 14, 2010, 01:03:58 AM

Previous topic - Next topic

rapidset

Hi,
  Installed the latest beta and had a runaway process, unfortunately the original process was not killed and the restart action happened repeatedly. I could kill the original offending process using process explorer without any issues, of course I also used process explorer to kill the tens of duplicate processes that had been spawned by process lasso

  It would be good to have some failsafe code so that in case of future non-bug issues that you dont fill up the process table if the original process doesnt die....

Log snippet
14-10-2010 13:51:29.093,VENTO,Rapidset,mum.exe,2684,Process was restarted due to watchdog,The process exceeded 45% of total CPU for 120 seconds (as configured).
14-10-2010 13:51:30.125,VENTO,Rapidset,mum.exe,2684,Process was restarted due to watchdog,The process exceeded 45% of total CPU for 120 seconds (as configured).
14-10-2010 13:51:31.140,VENTO,Rapidset,mum.exe,2684,Process was restarted due to watchdog,The process exceeded 45% of total CPU for 120 seconds (as configured).
14-10-2010 13:51:32.156,VENTO,Rapidset,mum.exe,2684,Process was restarted due to watchdog,The process exceeded 45% of total CPU for 120 seconds (as configured).
14-10-2010 13:51:33.171,VENTO,Rapidset,mum.exe,2684,Process was restarted due to watchdog,The process exceeded 45% of total CPU for 120 seconds (as configured).
14-10-2010 13:51:34.187,VENTO,Rapidset,mum.exe,2684,Process was restarted due to watchdog,The process exceeded 45% of total CPU for 120 seconds (as configured).
14-10-2010 13:51:35.218,VENTO,Rapidset,mum.exe,2684,Process was restarted due to watchdog,The process exceeded 45% of total CPU for 120 seconds (as configured).
..



Jeremy Collake

#1
Thanks for reporting this. I also saw this anomaly here and have it on my list. I'll make it a priority.

You are right, safety code should have been there. I *do* have safety code, several levels of it. Public betas have most of the safety code enabled, but I simply did not have any code to catch this condition.

As for killing those duplicated processes PL erroneously launched, don't forget you can select multiple processes with Process Lasso and kill them all -- a feature rarely found.

I apologize for the bad experience with the beta and promise this will be addressed very soon.
Software Engineer. Bitsum LLC.

Jeremy Collake

Ok, I have now added a safety catch so this can't occur. Now I need to investigate more, but at least you should *never* see multiple restarts occur like that again (in next beta).
Software Engineer. Bitsum LLC.

rapidset

Thanks, moving to 3.99.38 beta now

Reporting bugs is the whole point of the beta process after all :)
It speaks volumes (in a good way) that I only had to create a forum account after using beta versions of the tool for this long

Jeremy Collake

#4
I *hope* .38 does well for you. Honestly, I still have work to do in that area. We've got RC1 coming and THAT build has everything cleaned from top to bottom. If you do see problems in .38, please do report though so I can make sure they don't slip by in my final work here. Thanks much for your continued input ;)
Software Engineer. Bitsum LLC.

Jeremy Collake

In extended testing, I've seen I still need to make improvements to this code. I just wanted to let you know that if it isn't working as you desire still, it will be by final. I'm really testing the hell out of this and will not release it until it is done. Not only does it have to all work right, but it must be at least as efficient as v3 -- if not more efficient.
Software Engineer. Bitsum LLC.