Lockup of Everex Orgasmatron every couple of hours

[cj6814]

I've got the Everex Walmart special with the original Orgasmatron (Nerd Vittles for President!!!). Tried to follow the directions to the tee as I pride myself as only being an advanced novice.

I've had it running for 3 weeks or so in a small business environment. Very low call load because we haven't given up Ma Bell due to the instability of the PBX system.

1) Switched ISP to put the PBX box on a fixed IP outside of our LAN.
2) PBX Box on UPS to make sure no power issues are causing issues.
3) Update-scripts, update-fixes, update-source, and update-fixes again.

Info below so you know what is running

5 Extensions:
3 softphones
1 Aastra 57i - setup t*f*t*p according to Orgasmatron directions
1 Linksys SPA with two lines connected

Trunks are vitelity inbound and outbound

Status Version 1.2.6 released on Date 100708
pbx.dyndns.org on xxx.xxx.xxx.xxx (hidden for safety)
********************************************************************
* PBX in a Flash Version 1.2 Daemon Status
********************************************************************
* Running Asterisk Version : Asterisk 1.4.19.2
* Asterisk Source Version : 1.4.19.2
* Zaptel Source Version : 1.4.10.1
* Libpri Source Version : 1.4.4
* Addons Source Version : 1.4.6
********************************************************************
CentOS release 5 (Final) - 32 Bit ** Kernel: 2.6.18-53.1.14.el5
********************************************************************
For help on PBX commands than you can run type help-pbx *
********************************************************************

The system won't stay up for more than 3 days at a time but usually only 2-3 hours. The lockup is the whole thing....we lose phones...no GUI...and can't ssh. Hard reboot is required and a couple of times it has required some fsck to get it to boot back up.

Things I've observed:

1) 100% CPU usage at times.

If I ssh and run "top" to see what processes are going I've seen the user "root" with the command "bzip2" running 99.3% of the cpu. This is not always.

This isn't always what preceeds the lockup but has a couple of times when I've been on top of the situation and looking for the issue. If the 100% cpu usage starts it will always result in a lockup.

2) Action on the log that immediately preceeds lockup.

I found some other posts that ask about the following line:
Parsing '/etc/asterisk/manager.conf': [Oct 27 22:05:53] VERBOSE[4779] logger.c: Found

In my case this Parsing runs every two seconds until the whole system comes crashing down.

I can reboot the system and the parsing starts up within the 10 minutes or so. However, to try something new I rebooted the system (with "shutdown -r now") and then stopped asterisk for an hour and then restarted. The Parsing then did not show up on the logs for almost a day....however...when it did show up it ended up taking the system down again (I will attach a copy of the log file if it interests you. The lockup was between the 27th and 28th if you're following the time stamps).

Anyhow, I'm wondering if I should just do a fresh install as I haven't really found any posts that lead me to believe that this is a common issue. But, I'm not convinced that a fresh install would leave me with a different outcome.

Any information would be greatly appreciated as I find myself all of a sudden a voip nerd.

If I haven't included some vitel information let me know and I'll supply it.

Thanks in advance.

 

[thunderheart]

One thing I might try

Reformat and install a typical Linux distribution. Then run mprime to load the hardware up for say 24 hours. If mprime runs for 24 hours error free, you can pretty much rule out hardware instability. I like to do this on a new machine and periodically run sensors (from the LM Sensors package) over the test period to get an idea of the thermal performance. The mprime program puts a very heavy load on the machine so it will run pretty much as hot as it is ever going to. It may not be the best possible stress test but it is pretty darn good.

Dallas

 

[lopaka]

Stress test utilities

There are a few stress testing utilities that will run from a bootable CD that may save you the hassle of reformatting.

 

[wardmundy]

I've had Everex machines do this twice. The easiest solution is to get a flash drive (4GB should do it), install Tom's disk backup script, and make a full backup. Then burn the backup ISOs to CDs, and restore the system. Linux doesn't handle bad disk sectors like it should. If it hits one, this will cause it. Doing a full restore with Mondo seems to do a better job of blocking out bad sectors. Good luck!

 

[cj6814]

can the sip_nat.conf file hose my system?

Ward....thank you for the response. I decided to use this as an excuse to simply download and install the software for the Orgasmatron II. A clean start....no settings carried over.

Well....I did that...and all worked great for 3 days without a hick-up. HOWEVER...after 3 days up time I decided to update the sip_nat.conf file for external extensions (classic one way audio).

I thought this was very straight forward:

externip=my.fixed.ip.address
localnet=192.168.1.0/255.255.255.0

No less....and no more. The results were immediate...the system went down within 20 minutes...and continued to go down every 20 minutes.

Now when I changed the nat_sip.conf I did the "amportal restart" but I didn't reboot the entire system (this is full disclosure as I later read you're supposed to do this reboot as well). Upon testing my external extensions had full sound.

Thinking that this was the cause of my problems I decided to clear the nat_sip.conf file....."amportal restart"...."shutdown -r now" and see if the system would regain stability.

It now stays up several hours instead of 20-30minutes each time.

So....am I grasping at straws? Could this have thrown a wrench in the works or am I making too much out of random coincidence? I really want to duplicate one of the stable Walmart specials you've documented so well but I just can't seem to get it working.

Thanks in advance for any insight.

 

[cj6814]

see the post on memory

I posted my findings in a new thread since I think this info is very important for all the frustrated Walmart special owners that have lockups.

Thanks to all that listened to me talk out loud with these problems.