Odd problem with Linux since MT

RouterOS General
1

I set up my p2p shot, got my AP’s going, so now I have broadband, finally.

I have a Ubuntu 6.06 server running Squid and I’m in process of setting up a mail server with it. The problem is the Linux box will randomly just lag on the network. Not a ping lag, just won’t ping outside. I’ll ping google, I’ll get 3 pings in 30 seconds, none dropped, latency is 40ms or under, but it will only ping 3 times in thirty seconds. If I leave it, it will just stop pinging, then start back. I’ve updated Ubuntu, changed from OSPF to static routing.

The rest of the network runs fine, and I’ve even tested it with fresh install of Linux on my laptop, it does the same thing. MTU is 1500 on everything, and I’ve tested it at lower settings even though pinging with -l 1472 -f shows no fragmentation.

The only thing odd I’ve found is I cannot ping google by name from the AP’s, but the clients have no problems, aside from those running Linux.

These Linux boxes have run on high speed before without issue, I’ve duplicated the problem on multiple machines, and it only happens in my MT network. Even when I was on dial up running ICS, I could ping anything I wanted consistantly, with high latency of course, and no lag.

Any ideas guys?

2

0ldman -

(Hi Glenn) - after thinking about this at length…

Get MTR - Multi-trace Route (Ubuntu Universe)
Turn off proxy rules in your router
Turn off Squid…

Try MTR by name to google

If unsuccessful…what are your dns settings? How are they handled by your MT router? Do you use redirect so you can use the DNS caching feature of MT ROS or what do you do to handle DNS requests?

Version of ROS?

Now - you’ve only talked Ubuntu..do you have other Linux ‘flavored’ boxes on your network? How do they work?

R/
Thom

3

I double checked my proxy rules in MT and found I had not disabled one of them. It may have caused a loop from my proxy server to my proxy server(why it only happened sometimes, dunno).

Disabled that and everything is cranking along nicely now. Still might not be the problem as it is inconsistent, but we’ll see.

4

Did not fix the problem.

When I change the srcnat public address it will start working for awhile, I can ping google by name from the routerboards, everything is great.

Is there a problem with my srcnat config?

 0   ;;; GW NAT
     chain=srcnat action=src-nat to-addresses=74.x.x.4 to-ports=0-65535 
     src-address=192.168.0.0/24
5

Still doing it. I have disabled all NAT and ran masquerade, disabled all mangle and firewall rules, gone from OSPF to static routing, nothing has changed.

Windows machines have a brief hesitation, Linux machines all but stop. I’m downloading a live Deb CD to see if it happens there as well.

6

Finally disabled Nstreme, seemed to fix it. After a day, no change.

it will ping google with average time of 35ms, ping 10 times in 30 seconds.

No one else has run into this?

7

I enabled allow remote requests in DNS, changed my DNS servers in my DHCP pool to the MT routers and its doing much better. The lag during pings in linux only happens every now and then, which I have noticed in the past.

4 days and counting.