nonexistent processes in topas

[trifo]

Hi!

I have some processes listed in topas which I can not find with ps. Can this situation be 'reset' somehow? Rebooting the machine is not an option for now.

Topas Monitor for host:    host                 EVENTS/QUEUES    FILE/TTY
Tue Jul 20 11:40:13 2004   Interval:  2         Cswitch    2398  Readch  2782.5K
                                                Syscall   29298  Writech  135.0K
Kernel   30.2   |#########                   |  Reads      1538  Rawin         0
User     17.7   |#####                       |  Writes      436  Ttyout      191
Wait     35.7   |##########                  |  Forks        17  Igets        39
Idle     16.2   |#####                       |  Execs        21  Namei      3676
                                                Runqueue    2.0  Dirblk      476
Interf   KBPS   I-Pack  O-Pack   KB-In  KB-Out  Waitqueue   4.5
en1      189.8   413.0   382.5    89.9    99.9
css0      28.9    56.5    60.5    19.7     9.2  PAGING           MEMORY
                                                Faults     5102  Real,MB    2047
Disk    Busy%     KBPS     TPS KB-Read KB-Writ  Steals        0  % Comp     78.9
hdisk1   96.0    446.7   108.0     0.0   446.7  PgspIn        0  % Noncomp  20.6
hdisk0   61.0    442.7   108.5     2.0   440.7  PgspOut       0  % Client   17.9
                                                PageIn        0
java     (144532  4.5% PgSp: 9.2mb app          PageOut     111  PAGING SPACE
java     (66060)  0.0% PgSp:83.3mb app          Sios        107  Size,MB    2048
java     (50064)  0.8% PgSp:344.4m: root                         % Used     34.2
amqzlaa0_(132566  1.5% PgSp: 0.9mb MQSLEKER                      % Free     65.7
topas    (108662  0.5% PgSp: 0.5mb root
amqzlaa0_(11812)  0.5% PgSp: 0.9mb MQSLEKER
amqcrsta_(131222  1.0% PgSp: 0.9mb MQSLEKER        Press "h" for help screen.
amqcrsta_(75196)  0.3% PgSp: 0.8mb MQSLEKER        Press "q" to quit program.

PID 144532, 66060 and some more is not listed when I run "ps -ef"

--Trifo

 

[KenCunningham]

Maybe these are just transient processes which disappear between closing topas and running ps? That would be my initial guess.

 

[trifo]

Nope. These processes seem to exist in topas list for 2 days at least.

--Trifo

 

[MoshiachNow]

And if you run "ps -ef|grep java" you still won't see them ?

Long live king Moshiach !

http://www.7for70.com/

 

[trifo]

Well, now you see my problem. Do you have any clue what should I do?

--Trifo

 

[hfaix]

Is the PID always the same?

Is Java functioning correctly?

 

[trifo]

PID 144532, 66060 are apparent continuously for 2 days at least, while we can not find them in the output of ps. (Yes, I know how to use grep)

_other_ java processes are in good health, workink properly.

kill command claims 'No such PID'

Any more idea?

--Trifo

 

[columb]

Have you tried ps -fu app to see if that helps

 

[trifo]

Coulomb: well, in any ways, I can not make ps to list some PID-s which are apparent in topas.

ps -ef creates a list of _ALL_ processes in the system ( except kernel threads and some similar) "ps -fu app" creates a subset of PID-s created ba "ps -ef"

Even though there seems to be no such process.

--Trifo

 

[columb]

I wasn't trying to teach myu grandmother to suck eggs - I personally find that the -u flag helps when I can't see the wood for the trees. Additionally looking closely at what else app is doing might give some clues as to where these invalid PIDs are coming from
BTW what does the 't' option in nmon show?

 

[d3vzero]

Hi Trifo,

Try using just "ps <PID#>".

eg. ps 66060

I'm sure it will pick it up ;-)

____________________
Sometimes it pays to stay in bed on Monday, rather than spending the rest of the week debuging Mondays code.

 

[d3vzero]

Sorry,

Had meant to say that the ps -e option excludes kernel processes. (I suspect that you are seeing a kernel proc in topas).

All the Best.

____________________
Sometimes it pays to stay in bed on Monday, rather than spending the rest of the week debuging Mondays code.

 

[trifo]

OK, let's change the original question:

how does topas create its own process list? And how topas's method of collecting PID-s differs from the method of ps?

I suspect some kind of malfunction of topas.

Columb: sorry if you found me rude...


--Trifo

 

[d3vzero]

Hi Trifo,

Topas is using a function similar to "ps -A".

If you replace your original "ps -ef" with "ps -A" I suspect your grep will pick it up.

Best Regards.

____________________
Sometimes it pays to stay in bed on Monday, rather than spending the rest of the week debuging Mondays code.

 

[columb]

I suspect you're right. If ps won't report the process and kill can't find it my guess is that it isn't in the process table. I wouldn't be the first time big blue had released dodgy code! I've had a quick look and topas isn't released under GPL (yet) so we can't dig too deep.
Have you got nmon? If so can it help identify the PID? Although 4.5% cpu isn't bringing your system to it's knees I'm sure your as curious as I am as to why you get this.

 

[d3vzero]

Appols Guys,

Did'nt see the failed kill. :-(

Very weird.

Are you running 5L ???

If you are, see if you can see the PID in the /proc directory.

Something else to try would be lsof (with the -P option).

Brgs.

____________________
Sometimes it pays to stay in bed on Monday, rather than spending the rest of the week debuging Mondays code.

 

[trifo]

d3vzero: I am running 4.3.3 on this node. Tried lsof, but it was unsuccessful either. lsof can not locate my ghost processes.

--Trifo

 

[d3vzero]

Hi Trifo,

Very Strange.

You could prove the existance of the PID using crash.

eg PID 66060 = 1020c (in Hex).

echo "proc" | crash | grep 1020c

If this does not list a process, then I doubt the actual existance of those PIDs displayed by topas.

If it does list a process them I'm suprised that ps cant see it at all.

Till now I would have said that there were no long running processes that could not be displayed with either "ps -A" or "ps <PID#>". How wrong could I be eh?

Very interesting situation to find yourself ;-)

____________________
Sometimes it pays to stay in bed on Monday, rather than spending the rest of the week debuging Mondays code.

 

[KenCunningham]

Interesting in the Chinese sense of 'I hope you have an interesting life?'?

 

[d3vzero]

:-D

Actually, it was more in the sense of 'I wish I had an interesting life.' ;-)

____________________
Sometimes it pays to stay in bed on Monday, rather than spending the rest of the week debuging Mondays code.