onstat -l when logging is set to "/dev/null"

Hello! I upgraded a few weeks ago Online from 5.05 to IDS 7.30. I do not care about continuous recoverability, so I have my LTAPEDEV set to /dev/null. The problem is that when I display logging state via onstat -l I see all my logical log files are "U" (Used) "B" (backed up) and 100% filled (after all logs were cycled once), whereas in Online 5.05 unused logs were "F" (free) 0% . The head of this circular queue can be identified with "C" flag (Current log). But I do not know how to determine the tail (oldest log being used). I use this (current log MINUS oldest log with transaction open) to estimate/measure size of transactions. Is there a way to determine the oldest log with open transaction when LTAPEDEV=/dev/null (IDS7.30UC2 HP_UX10.20) ? If such a problem is described somewhere in FAQ, please tell me where to find it (and sorry for the question). Peter

Look at the "log begin" column of the onstat -x report. Any non-zero values describe and active transaction which began in the log with that unique id. So the oldest, ie smallest, of these log ids us the value you are looking for. Art S. Kagel "Piotr Wróblewski" wrote: > > Hello! > > I upgraded a few weeks ago Online from 5.05 to IDS 7.30. > > I do not care about continuous recoverability, so I have > my LTAPEDEV set to /dev/null. > > The problem is that when I display logging state via > > onstat -l> > I see all my logical log files are "U" (Used) "B" (backed up) > and 100% filled (after all logs were cycled once), whereas in Online 5.05 > unused logs were "F" (free) 0% . > The head of this circular queue can be identified with "C" > flag (Current log). > But I do not know how to determine the tail (oldest log being > used). I use this (current log MINUS oldest log with transaction open) to > estimate/measure size of transactions. > > Is there a way to determine the oldest log with open transaction > when LTAPEDEV=/dev/null (IDS7.30UC2 HP_UX10.20) ? > > If such a problem is described somewhere in FAQ, please tell me where to > find it (and sorry for the question). > > Peter