When running a Policy Server, this one might becomes unresponsive when
the Profiler gets activated with all components and all data.
Buffertracing registry has been configured the sm.registry.
Policy Server 12.8SP2 on RedHat 6
From pstack, we see 93 % of the threads are waiting for a lock or a
condition.
We see on pstack the following function in execution :
accept () from /lib64/libpthread.so.0
recv () from /lib64/libpthread.so.0
CSmFileOutput::Log(CSmErrLogMessage&) () from /opt/CA/siteminder/lib/libsmerrlog.so
open64 () from /lib64/libpthread.so.0
pause () from /lib64/libpthread.so.0
pkgapp.backtrace-1
79 threads in that state :
in __lll_lock_wait () from /lib64/libpthread.so.0
39 threads in that state :
in pthread_cond[email protected]@GLIBC_2.3.2 () from /lib64/libpthread.so.0
31 threads in that state :
[email protected]@GLIBC_2.3.2 () from /lib64/libpthread.so.0
6 threads in that state :
nanosleep () from /lib64/libc.so.6
1 thread in that state :
accept () from /lib64/libpthread.so.0
1 thread in that state :
recv () from /lib64/libpthread.so.0
1 thread in that state :
CSmFileOutput::Log(CSmErrLogMessage&) () from /opt/CA/siteminder/lib/libsmerrlog.so
1 thread in that state :
open64 () from /lib64/libpthread.so.0
1 thread in that state :
pause () from /lib64/libpthread.so.0
pkgapp.backtrace-2
97 threads in that state :
[email protected]@GLIBC_2.3.2 () from /lib64/libpthread.so.0
31 threads in that state :
[email protected]@GLIBC_2.3.2 () from /lib64/libpthread.so.0
21 threads in that state :
__lll_lock_wait () from /lib64/libpthread.so.0
6 threads in that state :
nanosleep () from /lib64/libc.so.6
1 thread in that state :
CSmFileOutput::Log(CSmErrLogMessage&) () from /opt/CA/siteminder/lib/libsmerrlog.so
1 thread in that state :
accept () from /lib64/libpthread.so.0
1 thread in that state :
recv () from /lib64/libpthread.so.0
1 thread in that state :
open64 () from /lib64/libpthread.so.0
1 thread in that state :
pause () from /lib64/libpthread.so.0
Upgrade to Policy Server 12.8SP4.
In case of Buffered tracing, A Ring buffer is introduced which uses a
size of 10 mb. Due to large size of data pumped in its causing issue.
Increase the size of RingBuffer to 30 or more :
Registry name : SizeOfRingBuffer
Path of the registry :
HKEY_LOCAL_MACHINE\SOFTWARE\Netegrity\SiteMinder\CurrentVersion\LogConfig
Current value is 10.