When I run truss -c -p <weblogic app server> process, I have the following:

faults -------------
FLTBOUNDS        246
total:           246

signals ------------
SIGSEGV          246
total:           246

syscall               seconds   calls  errors
read                     .105    3767     11
write                    .262    6446
open                     .038    1047      1
close                    .036    2085   1026
stat                     .005     217    199
lseek                    .000      12
fstat                    .000       6
pipe                     .000       1
ioctl                    .000       4      1
fdsync                   .001       8
fcntl                    .000      28
lwp_park                 .132    4409     91
lwp_unpark               .126    4672
poll                     .130    4796
sigprocmask              .032     713
setcontext               .002     246
setustack                .000      38
mmap                     .017      61
mprotect                 .003     126
munmap                   .000       8
getrlimit                .000       1
memcntl                  .000       6
yield                    .001      48
lwp_create               .005      38
lwp_exit                 .000      23
lwp_suspend              .002      71
lwp_continue             .002     107
lwp_kill                 .000      23     23
llseek                   .052    4324
door_info                .000       1
door_call                .000       1
lwp_schedctl             .000      38
resolvepath              .008     191
getdents64               .000      40
stat64                   .403    9826   8456
fstat64                  .041    2482
so_socket                .001       8
bind                     .000       1
listen                   .000       2
accept                   .001       5
connect                  .001       6      4
shutdown                 .000       1
send                     .000       3
getsockname              .000       2
getsockopt               .000       2
setsockopt               .000      10
                     --------  ------   ----
sys totals:             1.426   45950   9812
usr time:              29.374
elapsed:               83.220


Question: 
1. what lwp_park and lwp_unpark supposed to do
2. what is lwp_park w/ ETIME error means in Java
3. why my 32-bit application have stat64 function call?
4. what is FLTBOUNDS means

Appreciate any assistance in pointing me to the right track. Thanks.

BTW, I am running Solaris 9, Java 1.3.1_08, Weblogic 7.0
--Chi Hung @ Singapore

chihung@mbox2.singnet.com.sg (Chi Hung Chan) writes:

>When I run truss -c -p <weblogic app server> process, I have the following:

These two are related: you got 246 SIGSEGV each with the same faultcode
(FLTBOUNDS)

>faults -------------
>FLTBOUNDS        246
>total:           246

>signals ------------
>SIGSEGV          246
>total:           246

These two are used in the implementation of the threads library;
sometimes lwps need to sleep in the kernel and that's what lwp_park
does.

>lwp_park                 .132    4409     91
>lwp_unpark               .126    4672


>Question: 
>1. what lwp_park and lwp_unpark supposed to do
>2. what is lwp_park w/ ETIME error means in Java

Porbably a condition which timed out; really not interesting.

>3. why my 32-bit application have stat64 function call?

So it can use large files.

>4. what is FLTBOUNDS means

>Appreciate any assistance in pointing me to the right track. Thanks.


What is the problem you'd like to solve?

Casper
-- 
Expressed in this posting are my opinions.  They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.

Thanks for the reply.

The application is running on SFV480 w/ 14GB, but the Weblogic 7.0 is
certified to run JVM 1.3.1_08 which does not have a 64-bit
implementation. I am not able to set the min/max heap size of the JVM
to anything more than 4 GB. If this is the case, why I still have
stat64 function call w/ 90% of these calls resulted in error.

It seems that the timeout of lwp_park and lwp_unpark is pretty high.
Is there any way to give a higher value timeout. I cannot find any man
page on my Solaris 9 installation. Also, very little info in the web
too.

This exercise is not trying to solve any of the performance problem,
cos the performance in likely to be on the database side. Just for a
better understanding. Cheers.
--Chi Hung

Casper H.S. Dik <Casper.Dik@Sun.COM> wrote in message news:<3ffe79ac$0$326$e4fe514c@news.xs4all.nl>...
> chihung@mbox2.singnet.com.sg (Chi Hung Chan) writes:
> 
> >When I run truss -c -p <weblogic app server> process, I have the following:
> 
> These two are related: you got 246 SIGSEGV each with the same faultcode
> (FLTBOUNDS)
> 
> >faults -------------
> >FLTBOUNDS        246
> >total:           246
>  
> >signals ------------
> >SIGSEGV          246
> >total:           246
> 
> These two are used in the implementation of the threads library;
> sometimes lwps need to sleep in the kernel and that's what lwp_park
> does.
> 
> >lwp_park                 .132    4409     91
> >lwp_unpark               .126    4672
> 
> 
> >Question: 
> >1. what lwp_park and lwp_unpark supposed to do
> >2. what is lwp_park w/ ETIME error means in Java
> 
> Porbably a condition which timed out; really not interesting.
> 
> >3. why my 32-bit application have stat64 function call?
> 
> So it can use large files.
> 
> >4. what is FLTBOUNDS means
>  
> >Appreciate any assistance in pointing me to the right track. Thanks.
> 
> 
> What is the problem you'd like to solve?
> 
> Casper