Re: more about kill -9, tbmode -z ...

In article <2nsng6$5vs@emory.mathcs.emory.edu>, ccmp@dec1.tvp.com.pl (Michal Pegierski) says: > >Hi, > > Some more questions and doubts about kill -9, tbmode -z etc. > >In ONLINE Administration Guide (v.5.00) , Chapter 2 p.2-26,27 >there is a description which is unsufficient, at least for me. > >The questions are: >1. How tbmode -z pid acts ?; make a kill call (with what SIG), or > communicates through shared memory ?; because server proccess > and application process communicates through pipe could it be SIGPIPE ? I don't have source, but it seems to set a bit in shared memory that tells the relevant SQLTURBO process "it's time to die". The SQLTURBO process must check that flag periodically, and terminate itself normally. Most versions of OnLine (4.00, 4.10, early 5.00) don't check this flag during some operations (index build, singleton SQL select) and won't die for a long time. >2. How to kill safetly application process killing at the same time database > server (from keyboard Ctrl-C or interrupt can do it) ? The safest way is to kill the front end (ISQL, 4GL program, ESQL program) process with Ctrl-C or kill -(anything but 9). The back end SQLTURBO will die in its own time. If you really want to make the SQLTURBO go away now, an Informix analyst told me to use kill -13 or -15, but only after waiting several minutes after killing the front end. Jim Burks jburks@promus.com Database Administrator The Promus Companies Memphis, TN 38125 USA