Hi all,
I am relatively new to Canopy, however I have been involved with Tropos for some time.
We have a simple Canopy P2P using a BHM and a BHS which works. After having some problems I started looking at the statistics, and for the Ethernet statistics saw high levels of outerror counts and no carrier errors, both of which are the same value.
I created a test setup in the office and saw these errors again so started changing ethernet speeds and duplex. The errors stopped when the Canopys where set to 100Mb Half duplex.
The cables are short, about one meter. One Canopy is plugged into an unmanaged 8 port switch via a Motorola PoE. The other is plugged into a Tropos router providing PoE via a cross over cable. This is the same for the live installation too. I get errors on both Canopys.
Should I be able to use full duplex?
I am running 8.1.5-BH DES PLD version 9. I think the live units are running 8.1.4-BH DES.
TIA
Are they P10 units? If not try downgrading them to 7.1.4 and look at the results again.
They are all P10 units. The live units are running 8.1.4-DES and my test units are running 8.1.5-DES.
I ran some throughput tests as we have decided to run at 100TX-HD. These results are from my test setup, running nuttcp between to SUSE 10.2 machines. These figures are really only for comparison. Ethernet interface changed on both master and slave units:
100TX-HD 3.4720mb/s
100TX-FD 3.4634mb/s
10TX-HD 2.6680mb/s
10TX-FD 0.2340mb/s
I was looking for any significant impact when changing to 100TX-HD and found it marginally quicker. I doubt I will be running any more tests to see if the switch or Tropos router used in the test setup is causing the low throughput at 10TX. The Canopy’s have only a 10Mb/s link, 5mb/s HD, 3.5mb/s effective HD so didn’t expect such low figures with the ethernet interface set at 10TX.
Has anyone else run similar tests?
For anyone still following this thread, I found this in the 8.1.5.1 release notes:
On newer, P10 series hardware, every Ethernet packet transmitted
over the wired interface is resulting in a CarSenseLost error.
Functionality is not affected, but the statistic will report erroneously
high. CarSenseLost (Carrier Sense Lost) is shown on each
module on the Statistics > Ethernet page.
The change was made and the errors stopped. The web page appears to be more responsive too, but that could just be subjective.
The errors could be real. I had two gamers that had over 1M of errors at 100FD over a seven day period and complained of high latency while gaming. After unchecking the 100FD in the radio both of them said their latency was cut in half and the excessive errors on the ethernet link had stopped. Both radios are 5.7 P10. Both customers have routers, Linksys and Netgear.
Personal opinion is that the router manufacturers didn’t expect their equipment to be plugged into something that runs at 100FD on the WAN port (DSL and CABLE modems) because I have never seen any errors when going directly into a PC. Remember, each component in that router came from the lowest bidder…
Currently checking the ethernet stats on all my customers in my spare time and unchecking the 100FD box whenever I find one with excessive errors.
Most consumer routers are only 10Mbps on their WAN ports. If you wanted higher (100) you’d really have to do some shopping.
Does anyone have any specific consumer routers that they are noticing this on? We use mostly Linksys (WRT54G) but if a customer has a separate brand it’d be nice to know which models people are seeing this issue on.
It’s a crap shoot. I have peope with Linsys that don’t have any issues, others that do. Best bet is to check the ethernet stats 24 hours after an install and if you see excessive errors after a day, uncheck the 100FD box.e
I have a few customers that still get a few errors even at 10FD or 10HD, but 10 errors per million octets should be an acceptable limit in most cases.
Not only are the radios showing errors, but my customers are complaining about having to reboot their router/ radios 6 or 7 times daily. We have installed about 75 2.4 P10 radios and we have had to uncheck the 100MB FD on about half of them. These customers have had a mix from Netgear, Linksys, and Belkin routers and some have been straight into the PC. It would be nice if Moto would give us some input on this issue.