For a 5.2 machine, ps output is has incorrect time/date info if the uptime is over 249 days. Any process started has the date of the 249th day instead of the current time and processes started at boot have a time instead of a date. Here is the output from a box up 263 days: uptime: 11:31am up 263 days, 25 min, 14 users, load average: 0.00, 0.00, 0.00 date: Tue Nov 16 11:32:24 EST 1999 ps auxw: (xterm just started) root 2337 0.0 1.2 2436 1580 pb S Nov 2 0:00 xterm ps auxw: (for init) root 1 0.0 0.0 776 68 ? S 14:33 0:08 init The behavior is also on a box that has been up for 323 days.
Try upgrading procps -- see ftp://people.redhat.com/johnsonm/procps/
Unfortunately, I am unable to test this now. The boxes that were up that long have been rebooted. The box that was up 323 days made it to 385 (yay!) and then was reinstalled with 6.1. Currently, it is working on 180 days, so I should see if the bug is gone in 2 months. I will let you know. I am running procps 2.0.4. I see 2.0.7 is at the FTP site you sent. I will wait to upgrade the package until I reach ~249 days and then, if there is a problem, see if the update works. Damon
That did it! The 5.2 machine in question was running procps-1.2.9-2. I built the 2.0.7 binaries and it displayed the correct dates in ps. Thanks! (Sadly, it won't matter much longer since the machine is going to 6.2 today. :)