Re: Power problem leaves transaction lock hanging

> > But there may have been a parent of an 'sqlturbo' process which was > incorrectly attached to a PPID of 1 instead! If so, killing that > process, *NOT* the 'sqlturbo', may well have sorted the problem! Not incorrect at all. init (pid of 1) owns whatever connects you. That connection in turn owns the application which is running That application owns the sqlturbo. Let us say: PID PPID 1 0 init 9916 1 hpterm ....... 9917 9916 foo 9918 9917 sqlturbo jparker 4.10.UD2 ....... If hpterm goes away, then foo may not realize this and keep running. It can no longer have a PPID of 9916, because 9916 is no more. Therefore it goes back to the previous parent or 1 - init. This is correct behaviour. With users taught to ALT-F4 all over the place this is a common problem. We have a script which wakes up every 5 minutes or so and looks for processes which are owned by init which shouldn't be. Whenever it finds one it pops a window on my (or whoever is around) tube indicating that a potential runaway has been found. We then go out and check and ensure that indeed it IS a runaway then we kill it. (Then we send a nasty_gram to the perp - used to get 5 or 6 a day, now it's like 1 a month). Be sure that you kill foo and NOT the sqlturbo. The sqlturbo will die gracefully of its own accord when it realizes that it's parent is gone, but if you kill it when it's in a critical section... well anyway that's all covered in the FAQ. cheers j. _____________________________________________________________________________ Jack Parker - Hewlett Packard, DMD/IS Boise, Idaho, USA jparker@hpbs3645.boi.hp.com _____________________________________________________________________________ "I'm with the IRS, I'm here to help you" _____________________________________________________________________________ Any opinions expressed herein are my own and not those of my employers. _____________________________________________________________________________