08-22-2006
socket close hangs and CPU go to 100%
Hello,
I'm currently having a problem with HPUX.
The application is a C app. It's a socket server.
It runs mostly fine, but under some circumstances (I can not replicate it), the app hangs and the CPU goes to 100%.
I have use gdb to attach to the app, and it was doing a close().
the close() system call is the one that is hanging.
--------------------
(gdb) where
#0 0xc00000000031a5d0:0 in _close_sys+0x30 () from /usr/lib/hpux64/libc.so.1
#1 0xc00000000032ac40:0 in close+0x140 () from /usr/lib/hpux64/libc.so.1
#2 0x4000000000053950:0 in server (soc_id=5, client_address={sin_family = 2,
sin_port = 3149, sin_addr = {s_addr = 3489154679},
sin_zero = "\000\000\000\000\000\000\000"}) at src/tpswitch.c:982
#3 0x4000000000055b50:0 in main (argc=1, argv=0x9fffffffffffe8a0)
at src/tpswitch.c:1244
-----------------------
The previous stack trace shows that the app is in _close_sys+0x30() from the c library.
The descriptor passed to close() is a socket descriptor.
Thanks in advance for your Help!
9 More Discussions You Might Find Interesting
1. Programming
I have a SUN environment running an WebLogic that communicates w/a 3rd party running IIS. When the IIS site goes down (frequently), I am stuck with sockets in an ESTABLISHED state, and cannot seem to figure out how to avoid this. No exceptions are thrown as I can still open connections to the IIS... (1 Reply)
Discussion started by: teledelux
1 Replies
2. HP-UX
Hi all,
I have a HP-UX 11.23 that have a Server establishing connections on port 8888 .
The problem is that when i need to stop and restart the Server, the connections mantain the same state and i need to wait about 20-30 minutes before all connections finishes.
The connections remain at... (2 Replies)
Discussion started by: Renato Gregio
2 Replies
3. UNIX and Linux Applications
I have written a socker program. I have executed that program many times without closing the socket. So I want to find which all sockets binded with which file descriptor.
Is there any way to close those socket, which have been opened in that program's execution.
please help me!.. (3 Replies)
Discussion started by: pa.chidhambaram
3 Replies
4. Programming
Interesting issue. There was some discussion on the LKML last year regarding the potential problems in concurrent applications reusing file descriptors in various scenarios. The main issue is that the reuse of a file descriptor and reception of data in a threaded application can be confused pretty... (1 Reply)
Discussion started by: ramen_noodle
1 Replies
5. UNIX for Advanced & Expert Users
Hi,
I was porting ipv4 application to ipv6; i was done with TCP transports. Now i am facing problem with SCTp transport at runtime.
To test SCTP transport I am using following server and client socket programs. Server program runs fine, but client program fails giving Invalid Arguments for... (0 Replies)
Discussion started by: chandrutiptur
0 Replies
6. Solaris
Hello Friends,
On one of my Solaris 10 box, CPU usage shows 100% using "sar", "vmstat". However, it has 4 CPUs and prstat and glance are not showing enough processes to justify high CPU utilization.
=========================================================================
$ prstat -a
... (4 Replies)
Discussion started by: mahive
4 Replies
7. IP Networking
Is the last two line necessary?
#include <sys/types.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <arpa/inet.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
int main(void)
{
struct sockaddr_in stSockAddr;
... (0 Replies)
Discussion started by: vistastar
0 Replies
8. Programming
can someone provide an example, where if the parent process quits for any reason, then the child process will also close? (3 Replies)
Discussion started by: omega666
3 Replies
9. IP Networking
I need clarification on whether it is okay to set socket options on a listening socket
simultaneously when it is being used in an accept() call?
Following is the scenario:-
-- Task 1 - is executing in a loop - polling a listen socket, lets call it 'fd', (whose file descriptor is global)... (2 Replies)
Discussion started by: jake24
2 Replies
LEARN ABOUT CENTOS
btraceback
BTRACEBACK(1) Network backup, recovery and verification BTRACEBACK(1)
NAME
btraceback - wrapper script around gdb and bsmtp
SYNOPSIS
btraceback /path/to/binary pid
DESCRIPTION
btraceback is a wrapper shell script around the gdb debugger (or dbx on Solaris systems) and bsmtp, provided for debugging purposes.
USAGE
btraceback is called by the exception handlers of the Bacula daemons during a crash. It can also be called interactively to view the cur-
rent state of the threads belonging to a process, but this is not recommended unless you are trying to debug a problem (see below).
NOTES
In order to work properly, debugging symbols must be available to the debugger on the system, and gdb, or dbx (on Solaris systems) must be
available in the $PATH.
If the Director or Storage daemon runs under a non-root uid, you will probably need to be modify the btraceback script to elevate privi-
leges for the call to gdb/dbx, to ensure it has the proper permissions to debug when called by the daemon.
Although Bacula's use of btraceback within its exception handlers is always safe, manual or interactive use of btraceback is subject to the
same risks than live debugging of any program, which means it could cause Bacula to crash under rare and abnormal circumstances. Conse-
quently we do not recommend manual use of btraceback in production environments unless it is required for debugging a problem.
ENVIRONMENT
btracback relies on $PATH to find the debugger.
FILES
/usr/lib/bacula/btraceback
The script itself.
/usr/sbin/btraceback
symbolic link to /usr/lib/bacula/btraceback
/etc/bacula/scripts/btraceback.gdb
the GDB command batch used to output a stack trace
AUTHOR
This manual page was written by Lucas B. Cohen <lbc@members.fsf.org>
SEE ALSO
bsmtp(1)
Kern Sibbald 6 December 2009 BTRACEBACK(1)