Re: Orphaned sql turbo processes.

hugh@nezsdc.icl.co.nz (Hugh Grierson) writes: >In article <360t73$ej5@status> chrism@iconz.co.nz writes: >>I remember getting a notice from informix - actually in the Turbo (3.xx) >>days that warned that killing an active sqlturbo process could lead to >>corruption in the db. Specifically the tbmode -z option which (i happen to >>know) did a kill -15?? on the process. I'm also guilty of doing the same >>thing and it was ok but I was always nervous afterwards. >At least on our platform tbmode -z does a kill -16 (SIGUSR1). I can't >see that doing kill -16 yourself could cause any more or less damage >than a tbmode -z, dire warnings from others notwithstanding. I don't recall which kill tbmode -z does (-16 sounds right), but it is used to protect the database from the infamous kill -9. Whatever you use, it is probably OK, but why mess with it? Tbmode -z works and is supported. The problem with kill -9 is that it is (a) non-maskable, and (b) sqlturbo can't clean up after itself this way. When the DB is in a critical state, this *WILL* trash your database. One thing about tbmode -z: it can take some time to work, so be *patient.* If you have to clean up after some transactions I've seen out there, you'd take a few minutes, too! Thankfully this problem mostly goes away with DSA. David David Coburn coburn@informix.com Informix Software, Inc. ...uunet!infmx!coburn Bellevue, Washington (206) 451-9252 ========================================================================= Any opinions expressed here are not necessarily those of Informix (or for that matter, anybody else). =========================================================================