Problems with Slashdot.org/Linux/Lag

DaveBuchanan1337

Well-Known Member
Joined
Feb 12, 2005
Messages
254
Reaction score
0
DISCLAIMER: I have read the Readme.1st, and searched google for hours.

I got iBurst three days ago - and I have been struggling ever since. I live Approx. 1.5km Line-Of-Sight from the Bryanston Water Tower, and I have 5 green LED's on the box.

My router, which is connected to the rest of the lan via Wireless G (this has been tested with previous Dial-Up), is a smoothwall beta box, Kernel 2.4, intel 200mhz P1, 80mb ram, 1gb ancient seagate, 2x Realtek NIC's.

The router connects to UTD with supplied Cat 5 cable

This is the situation:

1. Jumpy pings / lost packets - pings alternate from 80ms to 1000ms to lost with no bandwidth load, just idling. I have checked all my XP machines for worms.

2. I can't access www.slashdot.org or certain other websites. A linux ping of star.slashdot.org results in Bad Code 10, which by googling I have determined means that my packets are being 'administrively filtered by host'. Other websites - say www.penny-acade.com are fine, but slow.

3. I have figured out the best MTU via method described in Readme.1st - which is 1350, and this is set on the box. No major improvements.

4. The best achieved d/l speed was 60KBytes/sec from a torrent tracker at torrent.duke.edu. This was also inconsistent. And only happened once.

5. I have tried connecting to iBurst with a laptop -> ethernet cable -> UTD desktop box, and have achieved same results.

I AM ANGRY - and out *out* of ideas

Can anyone suggest anything that I may have overlooked?
 
Sorry about my untidy thread,
Those are the links to newdotnet removal and what it does
 
There is 1 thing that you have overlooked, and I know this from personal experience, bcos I'm also forced to use the Bryanston Water Tower base-station. You have overlooked the facts: it is not a problem on your side; also you are in the same boat that I'm in. I know it doesn't help, but there you go, I madder than a rattle snake - especially after WBS took money off my bank account after sending me an email saying that they would only do so after resolving all the issues. WBS you deserve to go bankrupt :mad:
 
2 things here, firstly your MTU, (not my favorite) but it does affect things, check ic's sticky/siggi for more details on that one. My MTU however is set much lower than the recommended 1392, at 1392 I cannot access a large amount of sites including this one.

The other is that wonderful tower, once again ic will share tales of poor performance, I was on that tower during the week and I must admit its the worst of the lot.

imo, there is something seriously wrong with that tower. Ofcourse getting WBS to do something about it, well.... :rolleyes:
 
I've forgotten to update my siggy, my MTU is no longer 1392, I changed it about 01:00am (today) during the blackout to 1352 (after packet capturing using Ehthereal & not being able to get through to the helplessnessdesk, I saw something about 1352 in the packets received). Sadly changing to 1352 made no difference, and I already knew that the auth server had crashed, I was merely taking a chance thinking they'd been "testing 1352" again without telling us...

The result of 1352, definitely not an improvement at all.
 
ic said:
I've forgotten to update my siggy, my MTU is no longer 1392, I changed it about 01:00am (today) during the blackout to 1352 (after packet capturing using Ehthereal & not being able to get through to the helplessnessdesk, I saw something about 1352 in the packets received). Sadly changing to 1352 made no difference, and I already knew that the auth server had crashed, I was merely taking a chance thinking they'd been "testing 1352" again without telling us...

The result of 1352, definitely not an improvement at all.

Talking about teh auth server .. My login used to be [email protected] . After a few hours of zero helpdesk I tried just whatever and that is still working now. :confused: :confused:
 
Code:
>ping -f -l 1404 [url]www.iburst.co.za[/url]

Pinging [url]www.iburst.co.za[/url] [196.30.31.120] with 1404 bytes of data:

Reply from 196.30.31.120: bytes=1404 time=247ms TTL=126
Reply from 196.30.31.120: bytes=1404 time=296ms TTL=126
Reply from 196.30.31.120: bytes=1404 time=327ms TTL=126
Reply from 196.30.31.120: bytes=1404 time=296ms TTL=126

Ping statistics for 196.30.31.120:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 247ms, Maximum = 327ms, Average = 291ms

>ping -f -l 1405 [url]www.iburst.co.za[/url]

Pinging [url]www.iburst.co.za[/url] [196.30.31.120] with 1405 bytes of data:

Packet needs to be fragmented but DF set.
Packet needs to be fragmented but DF set.
Packet needs to be fragmented but DF set.
Packet needs to be fragmented but DF set.

Ping statistics for 196.30.31.120:
    Packets: Sent = 4, Received = 0, Lost = 4 (100% loss),

So as far as I can see MTU is still 1432.
 
Sounds a bit like a DNS issue though, and from what I saw this morning that is likely.
I'm sure there is definately some wierd firewall on the iBurst side that may be the cause of the packet loss, dns issues and my issues connecting to irc etc...
 
DaveBuchanan1337 - switch to IPCOP. I have been using either or for about 3 years now and each seems more applicable in some situations.

I was having connectivity issues with SmoothWall 2 BETA and switched to IPCOP 1.4. No more problems.

I find it stable and feature rich. There is also more support on the IPCOP forums than at SW.

It has taken care of the MTU issues, the squid management is great. Snort and firewall are easy to configure. It is a breeze.

The reason I prefer a dedicated firewall distro rather than rolling my own is that I don't feel like missioning to maintain it. Once setup; it just stays up.

After hardware failures (yes I am using it on old dodgy PI's with old hardware) the restore config has worked great and I have got my firewall back up and running in less than a hour. That has been on both SW and IPCOP.
 
DFantom said:
Sounds a bit like a DNS issue though, and from what I saw this
morning that is likely.
Tried changing the DNS last night and it fixed the intl problems first time.
The DNS iBurst provides by default are UUNet/IAfrica's ones.

The other thing is UUNet was having some intl problem yesterday and the day before which may be the cause of the speed issues.
http://www.uunet.co.za/_sanetwork/networknotices/networknotices.asp?txtAction=D&id=1397
http://www.uunet.co.za/_sanetwork/networknotices/networknotices.asp?txtAction=D&id=1398
http://www.uunet.co.za/_sanetwork/networknotices/networknotices.asp?txtAction=D&id=1399
 
Top
Sign up to the MyBroadband newsletter
X