First of all, I should say that I’ve been running Asterisk on a Fedora
Core 3 box since last May, but decided to do a reinstallation of
everything due to some problems we’ve had with echos during
conversations (100% SIP based, so no ZAP echos). We are talking about a
low-volume installation running off an Intel Celeron 2.53 GHz box, 512
MB RAM and some no-name MB
(https://www.legendmemory.com/modules.php?op=modload&name=News&file=article&sid=412
<https://www.legendmemory.com/modules.php?op=modload&name=News&file=article&sid=412>).
The internet connection is DSL based, no NAT, eight IP addresses.
The problems continued. I did some tests on the DSL line, which itself
seemed fine. But when I connected Asterisk to the network, suddenly ping
became unstable. We are talking about ping varying between 20 ms and 600
ms over a connection that usually should result in 20 ms ping. So I
decided to eliminate the chance of increased latency due to traffic,
bypassed the switch and hooked the Asterisk box up directly to the
router. I even disabled Asterisk itself. Still, the problem persisted.
It’s worth mentioning that the latency to the router itself was fine all
the time, we’re just referring to the latency of the Asterisk box.
So I started installing [EMAIL PROTECTED] on a spare computer, about the same
specs, only a different motherboard
(http://www.dealtime.com/xPF-Asrock_MB_PM800_ASROCK_P4VM800_RTL_P4VM800).
Same issues. At this very moment, I was left with a suspicion that there
might be something with the DSL network, so I contacted the DSL provider
and had them run tests on the line, which all came out just fine. I
switched DSL modem/router, still same problems.
I then decided to do installations of different Linux distros (no
Asterisk installations) on the spare computer, and did some interesting
observings:
- Debian: Better, only occasional (a few percentage of the pings
performed were above normal, and then only 200 ms above normal, compared
to Centos/Asterisk where 30-40% were above normal). Still not 100%
- Mandriva: Perfect latency all the way
- Centos (base installation): Problems equal to the ones described earlier
I did some further testing, and on a P4 3,0 GHz, 512 MB RAM and a
motherboard which I don’t know the maker /chipet, the CentOS
installation came out with just a small percentage lag in latency.
I have checked for IRQ errors, and there seems to be no conflicts. I do
see the network sharing IRQ with the USB bus, but this is common on
motherboards with everything integrated.
I would assume this can be defined as off topic, as it clearly does not
relate to Asterisk in particular. However, I am writing this in the hope
of that someone might have had similar issues before and possibly been
able to solve the problem. Google has not given me much on the subject.
It’s not that I don’t mind putting together a new box with separate and
higher level of quality components, but I’d much rather learn something
from this experience than giving up ;)
You've provide a significant amount of general information, but nothing
specific enough to guess at what might be going on.
Some things you might want to consider or try (at least to eliminate
possibilities) include:
1. run Ethereal and trace the packets to/from your asterisk box, looking
for anomalies in response times, layer-3 routing issues, etc.
2. run 'netstat -rn' to ensure the path from asterisk to the default
gateway is exactly what you expect.
3. ensure the duplex setting of the nic card matches whatever that is
connected to (eg, switch, hub). Don't depending on auto-negotiation
functions working correctly.
4. run 'top' to ensure some other process isn't consuming processor
cycles unexpectedly.
_______________________________________________
--Bandwidth and Colocation provided by Easynews.com --
Asterisk-Users mailing list
To UNSUBSCRIBE or update options visit:
http://lists.digium.com/mailman/listinfo/asterisk-users
