Is it the router?

BTW, I'm doing this testing on a wired desktop.

See "Wireless router locks up and has to be restarted" in Fast Fixes below.

Thanks John. I'm getting in touch with Linksys and will ask about the beta firmware but I'm current as of right now. Since the latest firmware update is already 2 years old, I would assume it's pretty stable.

No P2P activity currently on my network.

I'm guessing it's something else.

Thanks aga> [POSTED TO alt.internet.wireless - REPLY ON USENET PLEASE]

Given the age of the router, it simply may not have sufficient horsepower for higher performance.

I meant age as in design age (as indicated by latest firmware being about two years old), not necessarily unit age. Some older designs have difficulty dealing with high throughput rates.

Right, understood. I'll be sure to post here what I learn, if anything, from Linksys.

David

John Navas wrote:

would the same thing apply if the router was dropping clients after a couple hours use?

John Navas wrote:

My guess in that case would be ARP table overflow, another not uncommon problem with older routers.

uhhh......

any information (links, etc....) on how i mite fix that?

John Navas wrote:

1. Newer firmware, if available.
2. Better/newer product.

So, at Linksys support's suggestion, I connected directly to the modem for most of the day and didn't ever see the speed decline. I then did the following:

'On the Setup tab (in the router's settings), look for MTU, select enable and set the size to 1492, 1400 or 1300. Make sure to save the settings'

I chose 1492 (bigger's better, right?). So far, so good.

David

John Navas wrote:

snipped-for-privacy@gmail.com wrote: [MTU]

I've always used 1492, and never had a problem, but strictly speaking, bigger isn't always better. You want the largest MTU that will not cause fragmentation elsewhere, so sometimes smaller is better...

Stable or abandoned. It's hard to tell. Some products have known issues with no updates, even though they are still being sold new (SMC is dreadful for that).

Can you look at logs for the cable modem? What kind is it? http://192.168.100.1/logs.html on my Motorola, http://192.168.100.1/logdata.html on my RCA. http://192.168.100.1/eventlog_page.html on a Thomson. http://192.168.100.1 might give a menu, might not.

These don't show speed, per se, but they do show reboots, and what turned out to be the real indicator for me, T3 and T4 timeouts. I had a bad cable connection in a grounding block at the power pole on my roof. Techs liked to change the connectors when they came out, but it was a while before someone went up on the roof and changed the ones that were bad. That was over a year ago, I think. They swapped modems a couple of times, because they couldn't think of anything else to do. Basically, the repair techs are cable TV guys. If the signal is good on channel 2 and 200 at the moment that they look, everything must be fine. ;-) Sometimes they would look at the signal level being reported by the modem, so you might check your browser hsitory to see what pages they visited, but they never look at the logs.

I just looked at a Thomson. 4925 T-3 timeouts between 3/17 and 5/20. Lots of errors last seen on 5/11. I wasn't here at the time, but there might have been some observable issues. There was a power cycle on 5/11.

When I was having trouble, I would visit this page regularly, and make note of the counts, so I could see the trend. The RCA is sneaky. It only shows the last one of a series of repeated messages, so looking at that one every five minutes in a script was much more useful than a yearly glance.

I was actually just joking a bit with the 'bigger is better' remark.

So, it seems the Linksys recommendation was easy and effective. I guess the router was fine all along!

Thanks for all the help, David

William P.N. Smith wrote:

Here's another question: If I'm now getting speeds of 9+ mps on my wired desktop, why am I only getting ~3.5 mps on my wireless laptop?

Thanks, David

snipped-for-privacy@gmail.com wrote:

You've bought into the marketing hype of WiFi. _Half_ of 54 is the _best_ you can do under laboratory conditions. In the real world, 3.5 isn't awful.