Package: procps Version: 0.97-4
------------------------------------------------------------ valour$ w 5:07pm up 15 days, 20:31, 7 users, load average: 0.16, 0.11, 0.09, 3/76 User tty login@ idle JCPU PCPU what and1000 ttyp0 4:15pm 49 24 -bash (bash) and1000 ttyp1 4:15pm 30 2 bash jbk1000 ttyp4 4:41pm 11 2:12 58 xpilot mphhpd.ph.man.ac.uk ijackson ttyp5 5:04pm 22 /bin/bash ajm46 ttyp6 4:45pm 2:10 -bash (bash) ajm46 ttyp7 4:45pm 3 2 rlogin chiark.chu ajm46 ttyp8 4:45pm 18 1 1 rlogin thor valour$ date Fri Dec 1 17:07:31 GMT 1995 ------------------------------------------------------------ Notice that: (1) ijackson logged in at 5.04pm, but is allegedly idling 22 mins - this is clearly impossible, since it's now 5.07pm! (2) the same thing is true of ajm46, on ttyp6. This is a reasonably reproducable situation. The idle time is wrong from the appearance of the xterm, and is correct once something has been typed at it. Could the maintainer please have a look at how idle times are calculated ? [I suspect that the idle time is taken as the atime on the /dev/ttyp?? devices, but maybe they're not being updated during the login sequence, so that the idle time is really the time since that tty was last used. Of course, as soon as the user uses the shell, the idle time is correct.] Austin