Kitz ADSL Broadband Information
adsl spacer  
Support this site
Home Broadband ISPs Tech Routers Wiki Forum
 
     
   Compare ISP   Rate your ISP
 
Please login or register.

Login with username, password and session length
Advanced search  

News:

Author Topic: MTU test  (Read 263 times)

Weaver

  • Addicted Kitizen
  • *****
  • Posts: 5396
  • Retd sw dev; A&A; 3 × 7km ADSL2; IPv6; Firebrick
MTU test
« on: June 27, 2018, 10:36:53 PM »

There used to be an online service that would check your MTU for you. Some code embedded in a web browser would generate messages of varying sizes which it would send to a server checking for fragmentation.

But it is now just broken. There is still a web server but it simply comes back with a server error. I wish someone could mend it as it used to be so useful. I saw there a working alternative ?


* UPDATE: It turns out that it is something in Apple’s software or in third-party software that is broken, not the website in question.
« Last Edit: June 28, 2018, 01:31:58 AM by Weaver »
Logged

burakkucat

  • Global Moderator
  • Senior Kitizen
  • *
  • Posts: 22849
  • Over the Rainbow Bridge
    • The ELRepo Project
Re: MTU test
« Reply #1 on: June 27, 2018, 11:32:43 PM »

I have just tried it and, for me, it does not error but exits gracefully stating that --

Quote
Sending 32 bytes to w.x.y.z  <-  NO REPLY!

We are unable to test this host or IP.
Most likely ICMP traffic to that host or IP is not allowed or the host or IP is invalid.

My VMG1312-B10A, the interface at the edge of my LAN, is configured not to respond to external ICMP Echo Requests.
Logged
:cat:  100% Linux and, previously, Unix. Co-founder of the ELRepo Project.

Please consider making a donation to support the running of this site.

RTouris

  • Just arrived
  • *
  • Posts: 4
Re: MTU test
« Reply #2 on: June 27, 2018, 11:35:03 PM »

Works for me. However do bear in mind that you have to allow for your gateway to respond to PING (ICMP) requests. This is widely supposed to be a security risk and therefore is usually disabled by default.

Here's the reply:

Sending 32 bytes to xx.xx.xx.xx  <-  not fragmented

...

Sending 1473 bytes to xx.xx.xx.xx  <-  FRAGMENTED!

Sending 1472 bytes to xx.xx.xx.xx  <-  not fragmented


From the tests we did, we can assume that 1472 bytes is the largest unfragmented packet
size. The MTU size would be 1500, made up from 1472 payload and 28 ICMP/IP Headers
and payload information.
The maximum MTU size for xx.xx.xx.xx is:    1500

NB: The Internet Protocol requires that hosts must be able to process IP datagrams of at least
576 bytes (for IPv4) or 1280 bytes (for IPv6)
Logged

Weaver

  • Addicted Kitizen
  • *****
  • Posts: 5396
  • Retd sw dev; A&A; 3 × 7km ADSL2; IPv6; Firebrick
Re: MTU test
« Reply #3 on: June 27, 2018, 11:36:02 PM »

I simply get a 404 and then a second error in trying to find the 404-handler document.

Could it be doing browser-sniffing or some such? I don’t understand how that would help matters though because it used to work for me, unless of course I am getting mixed up and it is a different site that I am now trying.
« Last Edit: June 27, 2018, 11:41:53 PM by Weaver »
Logged

burakkucat

  • Global Moderator
  • Senior Kitizen
  • *
  • Posts: 22849
  • Over the Rainbow Bridge
    • The ELRepo Project
Re: MTU test
« Reply #4 on: June 28, 2018, 12:06:26 AM »

The inner scientist insisted that I repeat the experiment with WAN side ICMP echo requests enabled. Here's the result --

Quote
Sending 32 bytes to w.x.y.z  <-  not fragmented

Sending 750 bytes to w.x.y.z  <-  not fragmented

Sending 1125 bytes to w.x.y.z  <-  not fragmented

Sending 1313 bytes to w.x.y.z  <-  not fragmented

Sending 1407 bytes to w.x.y.z  <-  not fragmented

Sending 1454 bytes to w.x.y.z  <-  not fragmented

Sending 1478 bytes to w.x.y.z  <-  FRAGMENTED!

Sending 1466 bytes to w.x.y.z  <-  not fragmented

Sending 1472 bytes to w.x.y.z  <-  not fragmented

Sending 1475 bytes to w.x.y.z  <-  FRAGMENTED!

Sending 1473 bytes to w.x.y.z  <-  FRAGMENTED!

Sending 1472 bytes to w.x.y.z  <-  not fragmented


From the tests we did, we can assume that 1472 bytes is the largest unfragmented packet
size. The MTU size would be 1500, made up from 1472 payload and 28 ICMP/IP Headers
and payload information.
The maximum MTU size for w.x.y.z is:    1500

NB: The Internet Protocol requires that hosts must be able to process IP datagrams of at least
576 bytes (for IPv4) or 1280 bytes (for IPv6)
Logged
:cat:  100% Linux and, previously, Unix. Co-founder of the ELRepo Project.

Please consider making a donation to support the running of this site.

j0hn

  • Kitizen
  • ****
  • Posts: 1577
Re: MTU test
« Reply #5 on: June 28, 2018, 12:24:16 AM »

From the Chrome browser for Android
Quote
This online MTU test is in BETA. Please let us know, using our contact form, if you find any irregularities.


Sending 32 bytes to ####  <-  not fragmented

Sending 750 bytes to ####   <-  not fragmented

Sending 1125 bytes to ####   <-  not fragmented

Sending 1313 bytes to ####   <-  not fragmented

Sending 1407 bytes to ####   <-  not fragmented

Sending 1454 bytes to ####   <-  not fragmented

Sending 1478 bytes to ####   <-  FRAGMENTED!

Sending 1466 bytes to ####   <-  not fragmented

Sending 1472 bytes to ####   <-  not fragmented

Sending 1475 bytes to ####   <-  FRAGMENTED!

Sending 1473 bytes to ####   <-  FRAGMENTED!

Sending 1472 bytes to ####   <-  not fragmented


From the tests we did, we can assume that 1472 bytes is the largest unfragmented packet
size. The MTU size would be 1500, made up from 1472 payload and 28 ICMP/IP Headers
and payload information.

The maximum MTU size for ####  is:    1500

NB: The Internet Protocol requires that hosts must be able to process IP datagrams of at least
576 bytes (for IPv4) or 1280 bytes (for IPv6)

working perfect.
Logged
BT FTTC 55/10 ECI now Huawei cab
Zyxel VMG1312-B10A bridge mode with 1508 MTU + Asus RT-AC68U running Asuswrt-Merlin

Weaver

  • Addicted Kitizen
  • *****
  • Posts: 5396
  • Retd sw dev; A&A; 3 × 7km ADSL2; IPv6; Firebrick
Re: MTU test
« Reply #6 on: June 28, 2018, 12:40:27 AM »

I have found it. It is some bug related to the use of a Safari ad-blocker. I ‘whitelisted’ the particular site, and got no improvements but to cut a long strange story short I found that if I disable the ad blocker completely then this fixes the problem.
Logged

banger

  • Reg Member
  • ***
  • Posts: 685
  • Uno comms 80/20
Re: MTU test
« Reply #7 on: June 28, 2018, 01:05:21 AM »

Works for me but my Asus router MTU wont go over 1492.
Logged
Tim
www.uno.net.uk & freenetname
Asus DSL-N55U and ZyXEL VMG1312-B10A Bridge on 80 Meg TTB Fibre

https://www.thinkbroadband.com/speedtest/1502566996147131655

Weaver

  • Addicted Kitizen
  • *****
  • Posts: 5396
  • Retd sw dev; A&A; 3 × 7km ADSL2; IPv6; Firebrick
Re: MTU test
« Reply #8 on: June 28, 2018, 01:12:44 AM »

Actually, no. There is something really weird going on here. I tried a different web-browser, iCab, and it still fails. Now I presume that iCab will not be using the ad-blocker because I would have thought that it is a plug-in specific to Safari.

There is an iPad app that allows you to inspect on http traffic (by way of installing a component that acts like an http proxy). I think that it was one of our own community who put me on to it - extremely useful on occasion. I discovered that firing this app up and letting it record traffic makes that website work! God only knows how!

I am wondering if this is a much deeper Apple bug of some kind in the light of this.
« Last Edit: June 28, 2018, 01:27:14 AM by Weaver »
Logged

j0hn

  • Kitizen
  • ****
  • Posts: 1577
Re: MTU test
« Reply #9 on: June 28, 2018, 09:19:38 AM »

I recall in another thread someone mentioned something about IPV6 breaking something on the MTU test.
I'm sure they fixed that though as my phone has IPV6 and works.
Logged
BT FTTC 55/10 ECI now Huawei cab
Zyxel VMG1312-B10A bridge mode with 1508 MTU + Asus RT-AC68U running Asuswrt-Merlin

johnson

  • Reg Member
  • ***
  • Posts: 341
Re: MTU test
« Reply #10 on: June 28, 2018, 09:25:17 AM »

I recall in another thread someone mentioned something about IPV6 breaking something on the MTU test.
I'm sure they fixed that though as my phone has IPV6 and works.

Yeah see:
https://forum.kitz.co.uk/index.php/topic,21273.msg369502.html#msg369502

I'd wager its something to do with IPV6 still.
Logged

Weaver

  • Addicted Kitizen
  • *****
  • Posts: 5396
  • Retd sw dev; A&A; 3 × 7km ADSL2; IPv6; Firebrick
Re: MTU test
« Reply #11 on: June 28, 2018, 10:24:08 AM »

I have had IPv6 for seven years or so and no problems with until recently so I am convinced it is some recently introduced iOS release. iOS 11.4 which I am running now seems to have been a real stinker and I much regret installing it. I should have done my homework beforehand. I will have to try and invent some kind of policy and be much more cautious about taking on new releases in future. Such a policy would involve waiting around for a good while so as to be able to hear the word on the street about a new release, and that does always take time.
Logged

spring

  • Reg Member
  • ***
  • Posts: 322
Re: MTU test
« Reply #12 on: June 28, 2018, 10:33:00 AM »

Windows 10 had been self destructing last month, leading to people even moving to Linux. Some of them had "turned off" windows update but obviously not hacked it so Windows 10 decided to ignore its own OS settings [why are they there to begin with?].

I'm on 7 and install the latest updates [among them is kernel] but always check to make sure they're something I want, to do that I check the KB. They are always wording things with maximum confusion but it's possible to tell when they're throwing blanks around, and Googling when in doubt always shows the clear answer. Quite a few updates were impossible to understand or had a disturbing fishiness in their description and after Googling turned out to be Spyware or other things I wouldn't want. Another thing to note is that while one month's updates are fine, the next could break the OS. So I tend to wait a few months especially when I've noticed yesterday how things are for 10 in the past month. Right now I'm up-to-date for May and it's running better than any past updates [as far as I noticed while using it]. There's essentially no drawback to waiting 1 month, 3 months or even 1 year, it always depends on how much the user is able to afford bad updates vs missing out on vulnerability fixes/improvements. It's also possible to select as many security fixes and as little software updates to avoid new adventures.
Here are the updates I didn't install:
Quote
https://support.microsoft.com/en-us/help/971033/description-of-the-update-for-windows-activation-technologies [spyware]
https://support.microsoft.com/en-us/help/3021917/update-to-windows-7-sp1-for-performance-improvements [spyware]
https://support.microsoft.com/en-us/help/3068708/update-for-customer-experience-and-diagnostic-telemetry [spyware]
https://support.microsoft.com/en-us/help/3080149/update-for-customer-experience-and-diagnostic-telemetry [spyware]
https://support.microsoft.com/en-us/help/3133977/bitlocker-can-t-encrypt-drives-because-of-service-crashes-in-svchost-e [I don't use bitlocker, and it has issues]
https://support.microsoft.com/en-us/help/3150513/latest-compatibility-definition-update-for-windows [spyware]
https://support.microsoft.com/en-us/help/3172605/july-2016-update-rollup-for-windows-7-sp1-and-windows-server-2008-r2-s [newer but worse windows update client]

Quote
Such a policy would involve waiting around for a good while so as to be able to hear the word on the street about a new release, and that does always take time.
Yes, as long as you can wait, 3 months should be enough Google results, as well as the company hopefully improving that release in those 3 months.
« Last Edit: June 28, 2018, 11:37:03 AM by spring »
Logged
No one knows what is the taste of the void.

Weaver

  • Addicted Kitizen
  • *****
  • Posts: 5396
  • Retd sw dev; A&A; 3 × 7km ADSL2; IPv6; Firebrick
Re: MTU test
« Reply #13 on: June 28, 2018, 11:52:14 AM »

A very good point Spring. We should perhaps all be a bit more restrained and resist the temptation to install things straight away.
Logged

spring

  • Reg Member
  • ***
  • Posts: 322
Re: MTU test
« Reply #14 on: June 28, 2018, 12:13:37 PM »

Well Firefox updates from 60.0.0 to 60.0.2 I install right away without checking what it's even about as I'm lazy, just feel a tiny bit uncomfortable in case they messed up :D, but that's normal.
Logged
No one knows what is the taste of the void.
 

anything