Bug may actually apply to the kernel package rather than the procps package. Output included below shows the problem. date shows the correct system time, however, "ps" appears to have been launched several minutes in the future. The whole command sequence completes immediately. Problem observed on many different sarge machines running 2.6 kernel. Could be a kernel issue reporting the wrong time somewhere in /proc but I'm not exactly sure how to capture that. stat /proc/PID shows the right ctime for the inode...i'm not actually sure where ps gets the START information from. root@sentinel:~# date;ps u Thu Mar 8 23:31:43 MST 2007 USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND ........ root 4800 0.0 0.1 2476 1424 pts/2 Ss 23:35 0:00 -bash root 5613 0.0 0.0 1956 744 pts/2 R+ 23:35 0:00 ps u
merge 408879 408879 thanks AKA bug #408879
Hi, This is a really old bug about ps start times. There has been a lot of changes with hardware, kernels and procps itself so I'm closing this bug. If it does still occur on current systems, we can reopen it or open another, but I'd say its fixed due to the update of those three things. - Craig