CACTID Issues - Windows [RESOLVED]

mcutting · Post by **mcutting** » Fri Jun 29, 2007 7:18 am

Hmmmm.......

"Message: Local - Cacti Poller Statistics [Time] went above threshold of 150 with 223.7861"

Looks like a PHP script error "Input expected". No errors since this so far...

One thing that is very annoying is that the poller is 10 seconds faster when run from CLI in comparison to the Windows Scheduler...

mcutting · Post by **mcutting** » Fri Jun 29, 2007 7:38 am

Ignore the above poller overrun for the moment.

I forgot to compile CACTID with "autoconf;./configure"

Will post updates as they happen.

mcutting · Post by **mcutting** » Fri Jun 29, 2007 8:47 am

Ok - an interesting twist:

Poller still overruns, but the processes terminate correctly, with nothing left hanging...

mcutting · Post by **mcutting** » Fri Jun 29, 2007 10:36 am

Unfortunately, this version of the poller does not work for me - there are too many poller overruns. I am going to go back to the previous "J" version I have.

Post by **TheWitness** » Fri Jun 29, 2007 9:41 pm

What log level are you running at? Jimmy Conner (cigamit) found an incorrect log message format. That would cause segfaults. Turn you logging to LOW and see what happens.

I will fix that tonight I think. Between residences at the moment.

Larry

Post by **TheWitness** » Fri Jun 29, 2007 10:01 pm

Also, what version of the MySQL source do you have?

Larry

mcutting · Post by **mcutting** » Sat Jun 30, 2007 6:26 am

Hi Larry,

Running at MEDIUM logging level. Have put the ThreadSafe CACTID back in, and set the logging to LOW as suggested. I compiled this using the mysql-4.1.22 source binaries

Does this make a difference ? Which version of the MySQL source should I use ?

Post by **TheWitness** » Sat Jun 30, 2007 6:57 am

Use the latest 5.0.41 as the thread safe modules were experimental in 4.x. Don't know about 4.1.

TheWitness

mcutting · Post by **mcutting** » Mon Jul 02, 2007 12:17 am

Hi Larry,

Tried lowering the logging level to LOW as suggested. Not much difference - multiple segfaults still (see poller graph). Also, there are multiple instances of :

06/30/2007 06:50:13 PM - PHPSVR: Poller[0] ERROR: Input Expected, Script Server Terminating
06/30/2007 06:50:13 PM - PHPSVR: Poller[0] ERROR: Input Expected, Script Server Terminating
06/30/2007 06:50:13 PM - PHPSVR: Poller[0] ERROR: Input Expected, Script Server Terminating
06/30/2007 06:50:13 PM - PHPSVR: Poller[0] ERROR: Input Expected, Script Server Terminating
06/30/2007 06:50:13 PM - PHPSVR: Poller[0] ERROR: Input Expected, Script Server Terminating
06/30/2007 06:50:12 PM - CACTID: Poller[0] FATAL: Cactid Encountered a Segmentation Fault (Cactid thread)

In the log. There is also a stacktrace:

Exception: STATUS_ACCESS_VIOLATION at eip=610F6023
eax=00000000 ebx=00000064 ecx=00000000 edx=1A3240C4 esi=0000006A edi=00000000
ebp=1A32409C esp=1A324064 program=c:\wamp\www\cacti\cactid\cactid.exe, pid 3312, thread unknown (0x1658)
cs=001B ds=0023 es=0023 fs=003B gs=0000 ss=0023
Stack trace:
Frame Function Args
1A32409C 610F6023 (1A32D008, 00000000, 1A3240C4, 610E5132)
1A3240EC 610E650F (1A32D008, 1A324134, 00000000, 00000000)
1A32410C 610E6561 (1A324134, 00000000, 00000FFF, 1A326234)
1A32627C 61092D88 (1A326294, 00000400, 0048B0DB, 1A326694)
1A326A9C 00402AE4 (0048DB0C, 00000000, 00000000, 1A32CE64)
1A326ABC 0040A34C (0000000B, 70697263, 47726F74, 70756F72)
1A326BAC 61017B60 (00000094, 0000EA60, 000000A4, 1A326BF4)
1A326CCC 610977FC (0000000C, 1A326D04, 0000001C, 00000000)
1A326DDC 61018C4C (1A326EC8, 1A32CE68, 1A326EE4, 1A326E9C)
1A326E00 7C9037BF (1A326EC8, 1A32CE68, 1A326EE4, 1A326E9C)
1A326EB0 7C90378B (00000000, 1A326EE4, 1A326EC8, 1A326EE4)
1A3271E8 7C90EAFA (1A32D008, 00000000, 1A327210, 610E5132)
1A327238 610E650F (1A32D008, 1A327280, 00000000, 00000000)
1A327258 610E6561 (1A327280, 00000000, 00000FFF, 1A329380)
1A3293C8 61092D88 (0048C828, 0000000C, 000000C0, 1A32A540)
1A32CD58 00406F72 (0000000C, 00000000, 00000000, 007E47C0)
End of stack trace (more stack frames may be present)

I have recompiled CACTID using the MySQL5.0.4 client libraries (./configure --without-server --enable-thread-safe-client), and have replaced the poller around 10 minutes ago. So far, so good - let's see what happens.

mcutting · Post by **mcutting** » Mon Jul 02, 2007 7:17 am

Since upgrading the MySQL client binaries to 5.0.41, the segfaults have stopped (I hope...!), and CACTID version "J" is extremely stable. It has been running since around 5am this morning with no problems whatsoever. I will continue to post updates in this topic...

Post by **TheWitness** » Mon Jul 02, 2007 9:55 am

I'll be checking things from time to time as I am still in the process of moving. Keep me posted.

Larry

mcutting · Post by **mcutting** » Mon Jul 02, 2007 9:56 am

Thanks Larry.

Still looking good - no segfaults.

Thanks for all your help.

Post by **TheWitness** » Mon Jul 02, 2007 9:59 am

MCutting,

In the mean time, do the following from the Cactid build directory:

gdb cactid
break *0x00402AE4
break *0x0040A34C

Let me know what lines number you get back. Just in case you were wondering:

Exception: STATUS_ACCESS_VIOLATION at eip=610F6023
eax=00000000 ebx=00000064 ecx=00000000 edx=1A3240C4 esi=0000006A edi=00000000
ebp=1A32409C esp=1A324064 program=c:\wamp\www\cacti\cactid\cactid.exe, pid 3312, thread unknown (0x1658)
cs=001B ds=0023 es=0023 fs=003B gs=0000 ss=0023
Stack trace:
Frame Function Args
1A32409C 610F6023 (1A32D008, 00000000, 1A3240C4, 610E5132)
1A3240EC 610E650F (1A32D008, 1A324134, 00000000, 00000000)
1A32410C 610E6561 (1A324134, 00000000, 00000FFF, 1A326234)
1A32627C 61092D88 (1A326294, 00000400, 0048B0DB, 1A326694)
1A326A9C 00402AE4 (0048DB0C, 00000000, 00000000, 1A32CE64)
1A326ABC 0040A34C (0000000B, 70697263, 47726F74, 70756F72)
1A326BAC 61017B60 (00000094, 0000EA60, 000000A4, 1A326BF4)
1A326CCC 610977FC (0000000C, 1A326D04, 0000001C, 00000000)
1A326DDC 61018C4C (1A326EC8, 1A32CE68, 1A326EE4, 1A326E9C)
1A326E00 7C9037BF (1A326EC8, 1A32CE68, 1A326EE4, 1A326E9C)
1A326EB0 7C90378B (00000000, 1A326EE4, 1A326EC8, 1A326EE4)
1A3271E8 7C90EAFA (1A32D008, 00000000, 1A327210, 610E5132)
1A327238 610E650F (1A32D008, 1A327280, 00000000, 00000000)
1A327258 610E6561 (1A327280, 00000000, 00000FFF, 1A329380)
1A3293C8 61092D88 (0048C828, 0000000C, 000000C0, 1A32A540)
1A32CD58 00406F72 (0000000C, 00000000, 00000000, 007E47C0)
End of stack trace (more stack frames may be present)

mcutting · Post by **mcutting** » Mon Jul 02, 2007 10:05 am

Hi Larry,

See output below

$ gdb cactid
GNU gdb 6.5.50.20060706-cvs (cygwin-special)
Copyright (C) 2006 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB. Type "show warranty" for details.
This GDB was configured as "i686-pc-cygwin"...
(gdb) break *0x00402AE4
Breakpoint 1 at 0x402ae4: file util.c, line 497.
(gdb) break *0x0040A34C
Breakpoint 2 at 0x40a34c: file error.c, line 67.
(gdb)

Regds

Post by **TheWitness** » Mon Jul 02, 2007 10:34 am

Add this one:

break *0x00406F72

very interesting output. Looks like a forked process was in the process of dying. That would either be a popen for a script server or a script...

However, it may not have been valid as it was before you compiled with the new libraries...

TheWitness

CACTID Issues - Windows [RESOLVED]

Who is online