Bug#216436: same problem with sorting %CPU, but worse

Ross Johnson Ross Johnson <ross@homemail.org>, 216436@bugs.debian.org
Sun, 29 Aug 2004 14:37:07 -0500


I observe the same problem.  It appears that the list is intended to be 
sorted by the parent CPU usage rather than the sum of the usages of 
parent and children.  I would prefer the latter, as suggested in the 
previous posting to this bug report.

I also find a sorting problem when I turn off the process dependency 
view (unlike the previous post).  I am running setiathome with nice -19, 
which was launched by cron under a different user ID.  When I open the 
system monitor (which was last left in a sorted-by-CPU state), 
setiathome never shows up at the top.  I can use "ps aux | grep seti", 
so I know setiathome is taking over 90% of the CPU, but the system 
monitor never shows it at the top of the list.  [Actually, ps tells me 
that setiathome is not getting any CPU when I first start the system 
monitor - this is because setiathome is nice -19.  But setiathome starts 
getting CPU again after the system monitor gets settles in].  When I 
scroll around to find setiathome in the system monitor, I can see the 
CPU utilization in the window, above 90% as expected.  However, 
immediately on the next refresh, setiathome goes to the top of the list 
where it belongs, so I have to scroll back to the top to see it.  And 
now, everything seems to be sorted correctly, until I open the system 
monitor again.

This is not an impatience problem.  I can wait 15 minutes while watching 
setiathome with ps and it never gets to the top of the system monitor 
list.  I can also launch a second system monitor after "fixing" the 
first one, and setiathome never shows on the top of list in the second 
instance.  So, at this point the two instances of system monitor 
disagree.  And, again when I scroll around to find setiathome in the 
second instance (after waiting a long time), it pops to the top at the 
next refresh.

This is repeatable on my machine.  Is it possible that the latter 
problem is in some way related to the former?  Or, is this a separate 
problem?  I recently upgraded to sarge and sarge-proposed-updates.

Ross