asterisk users - May 2006 - Snom firmwares suck <--additional datapoint to consider

We have a large install of 360's running rev 4.1 with zero problems. I did
another, smaller install couple weeks ago with 40 360's running rev 5.3. In
both cases, the install was identical, same Asterisk version, same dialplan,
everything the same except the differences were:

1. Different firmware rev
2. Different physical LAN

Guess what? On the smaller install, lockups and reboots. Emailing Snom, I
got this response: 

"The mentioned behaviour occured under certain
network conditions and should have been fixed with V5.5 which is -though its
Beta state- the last verified stable and reliable version:
snom360: http://fox.snom.com/download/snom360-5.5b-beta-SIP-j.bin
I would not recommend to downgrade since you would be loosing some
functionality"

Changing firmware revs did not help, so that left the LAN.

I looked long and hard at the LAN and it was basically narrowed down to the
switches. In this smaller install, several cheapo Dlink ($30) switches
de-aggregate a Cisco Catalyst switch. What I noticed was that any phone
plugged direcly into the Catalyst did *not* lock up or reboot. Any phone
plugged into the crap switches experienced the lockup. So now we are down to
the cheap switches themselves. We are nuking the Dlink switches and
replacing them with 3com workgroup switches, same as what we use in the
large install to good effect, and I fully expect the problem to dissapear. 

It's unfortunate that Snoms have a propensity to freak out in certain
environments but I don't think it would preclude me from using Snom in the
future. As long as one is aware of this issue, it should be easy enough to
work around.

> Changing firmware revs did not help, so that left the LAN.
>
> I looked long and hard at the LAN and it was basically narrowed down to the
> switches. In this smaller install, several cheapo Dlink ($30) switches
> de-aggregate a Cisco Catalyst switch. What I noticed was that any phone
> plugged direcly into the Catalyst did *not* lock up or reboot. Any phone
> plugged into the crap switches experienced the lockup. So now we are down
to
> the cheap switches themselves. We are nuking the Dlink switches and
> replacing them with 3com workgroup switches, same as what we use in the
> large install to good effect, and I fully expect the problem to dissapear.
>
> It's unfortunate that Snoms have a propensity to freak out in certain
> environments but I don't think it would preclude me from using Snom in
the
> future. As long as one is aware of this issue, it should be easy enough to
> work around.
Thanks for your input!

Previously I was using Nortel 10/100 switches, I replaced them some 
weeks ago with 3C16479 gbit switches. The phones are connected directly to 
the gbit switches. By coincidence I dit notice on one phone that in a 
split second a message appeared 'Ethernet cable disconnected'. Because I
have cable unplug set to ignore the conversation was not interrupted and 
the conversation could continue.

But that still doesn't solve the occasional lockup.

One phone was giving me *lots* more reboots than others but that was due 
to it running firmware 6.0.4 without having the ramdisk converted to jffs. 
Apparently the firmware didn't like that at all or just runs out of 
memory and decides to reboot.

> 
> I looked long and hard at the LAN and it was basically narrowed down to
the
> switches. In this smaller install, several cheapo Dlink ($30) switches
> de-aggregate a Cisco Catalyst switch. What I noticed was that any phone
> plugged direcly into the Catalyst did *not* lock up or reboot. Any phone
> plugged into the crap switches experienced the lockup. So now we are down
to
> the cheap switches themselves. We are nuking the Dlink switches and
> replacing them with 3com workgroup switches, same as what we use in the
> large install to good effect, and I fully expect the problem to dissapear.
We had the same problems with some cheap LevelOne Switches.
The Snoms rebooted during a call, calls dropped etc.
Replacing the switches was the solution.

Guido

> I looked long and hard at the LAN and it was basically narrowed down to the
> switches. In this smaller install, several cheapo Dlink ($30) switches
> de-aggregate a Cisco Catalyst switch. What I noticed was that any phone
> plugged direcly into the Catalyst did *not* lock up or reboot. Any phone
> plugged into the crap switches experienced the lockup. So now we are down
to
> the cheap switches themselves. We are nuking the Dlink switches and
> replacing them with 3com workgroup switches, same as what we use in the
> large install to good effect, and I fully expect the problem to dissapear. 
So does anyone have any theories as to what the technical difference 
between a "good" switch and a "bad" or "cheapo"
switch actually is?
Lower latency? Better grounding? More cowbell?

- Mike

Dr. Michael J. Chudobiak wrote:
>> Blaming the 3com switch is very likely to be the wrong root cause. 
>> High probability the 3com was not configured properly for the phone.
>
> Just curious - what configuration issues did you have in mind?
>
> - Mike
Replacing it with a Catalyst?
Andrew

>More cowbell?
Shit, you owe me a new keyboard! Funniest thing I've *ever* read on the
list. 

I've experienced the auto-negotiate issue with Snom's before. I forgot
to
mention that we make it part of our standard install to force 100baseT-full.
I've also noticed the Catalyst does the spanning-tree thing and waits up to
30 seconds before  enabling the port - this can cause problems with Snoms
because they boot before the Catalyst enables the port, causing registration
to fail. Then you warm-boot the Snom and everything's OK. 

One last interesting tidbit: We have a *lot* of Dell Dimensions with super
craptastic embedded Ethernet. They will auto negotiate with a Snom (plugged
into the PC port) to 100baseT full, but then you can't ping or TX past the
phone itself. Oddly enough, it gets an IP from our DHCP server OK. Forcing
the Dell to 100baseT full, half, or even 10 full works 100% of the time.
This never happens on any kind of decent Ethernet card like an 82557 chip or
3com. If we have an Optiplex, it *just works*

On Fri, 26 May 2006, Guido Hecken wrote:
> We had the same problems with some cheap LevelOne Switches.
> The Snoms rebooted during a call, calls dropped etc.
> Replacing the switches was the solution.
A switch should NEVER cause ANY device to lockup, ever. Period.
If a phone locks up / reboots due to something a switch sends, then the 
phone is faulty.

-Dan

> On Fri, 26 May 2006, Guido Hecken wrote:
> > We had the same problems with some cheap LevelOne Switches.
> > The Snoms rebooted during a call, calls dropped etc.
> > Replacing the switches was the solution.
> 
> A switch should NEVER cause ANY device to lockup, ever. Period.
> If a phone locks up / reboots due to something a switch sends, then the
> phone is faulty.
> 
Okay, it shouldn't reboot if not told to do so but a switch with e.g.
corrupt mac tables can bring your whole network down and the phone has no
chance too. 
However, if the SNOMS still reboot with the follwing settings attached, I
would also think of a possible bug in the firmware.
Setup/Advanced

Detect Ethernet Cable Unplug: off  
Action on Ethernet cable replug: ignore

Guido

On Thu, May 25, 2006 at 04:42:57PM -0600, Colin Anderson wrote:

CA> I looked long and hard at the LAN and it was basically narrowed down to
the
CA> switches. In this smaller install, several cheapo Dlink ($30) switches

What switches you mean? How they named?

-- 
JID: ds@im.seiros.ru
ICQ: 58417635 (please, use jabber, if you can)

http://freesource.info/