I have Tru64 4.0f running on an ES40. Occasionally and seemingly
randomly at times "ps auxwww" hangs the terminal in which I run it.
I have used the collect command and successfully listed out all
processes running. What I have noticed are several hundred "grep"
and "ps" commands running. *scratch head* I have no clue why this
would be the case nor what could be causing it. I am able to
easily kill the grep process with a simple cut | xargs line but the
ps commands refuse to die. This machine is not accessible to the
outside so I can virtually 100% rule out a hacked ps command or the
While I can work around ps * and the like I can not work around
the severe performance impact this is having on the system. The odd
part is I see hardly any processor activity (4 633 ev67 procs and
4GB of ram) and yet the system is running at least half the speed it
was before this occures.
To date the only "fix" I have found is a reboot. This is neither
practical nor doable considering this machine accounts for about 60%
of our production facility and must be up near 24/7. I have seen it
run for 4 days and do this and for 123 days only to ultimately
fall into this rutt which appears to have no easily noticeable
I have read a number of articles in which people explained a
similar issue. It was indicated this could have something to do
with streams. Others suggested tracing the PS command. The later
I do not have the knowledge to do and I am guess I do not have the
tools (my only experience running a trace is under linux). Still
others indicate it is fixed in 5.x yet one of the posts was with
reference to a 5.0 system having the same problem.
Could this be some hardware issue? Bad memory? Bad motherboard or
baseplane? I remember a time when the wrong memory was installed on
an alphaserver 1200 and it did strange things like crashes only once
X amount of memory was used but to date I have never seen anything
like this on any system wether alpha or tru64/osf or otherwise.
Anyone have a freaking clue whats going on with this machine?