Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Jan 24 13:06
    thomas-mangin commented #950
  • Jan 24 13:06
    thomas-mangin commented #950
  • Jan 24 00:17
    thomas-mangin commented #950
  • Jan 23 22:34
    cooperlees commented #950
  • Jan 23 22:33
    cooperlees commented #950
  • Jan 23 22:30
    adrian62 commented #950
  • Jan 23 22:19
    thomas-mangin commented #950
  • Jan 23 22:19
    thomas-mangin commented #950
  • Jan 23 22:11
    thomas-mangin commented #950
  • Jan 23 22:02
    thomas-mangin assigned #950
  • Jan 23 21:45
    adrian62 edited #950
  • Jan 23 21:39
    adrian62 commented #950
  • Jan 23 21:32
    cooperlees commented #950
  • Jan 23 21:25
    adrian62 opened #950
  • Jan 23 19:00

    vincentbernat on sid

    Merge tag '4.2.4' into debian/s… New upstream release d/patches: remove patch applied… and 4 more (compare)

  • Jan 23 18:37

    vincentbernat on sid

    d/lintian: ask lintian-brush to… (compare)

  • Jan 21 08:14
    topranks commented on 6e24499
  • Jan 21 01:25
    thomas-mangin commented on 6e24499
  • Jan 21 01:19

    thomas-mangin on master

    updating version to 4.2.4 (compare)

  • Jan 21 01:19

    thomas-mangin on 4.2.4

    updating version to 4.2.4 (compare)

Thomas Mangin
@thomas-mangin
feel free to grab me here
no issue
sanjmonkey
@sanjmonkey
1000x thankyous again
Thomas Mangin
@thomas-mangin
you are welcome :smile:
Thomas Mangin
@thomas-mangin
@sanjmonkey where is the code located ?
and good morning
Thomas Mangin
@thomas-mangin
sorry I am blind
Thomas Mangin
@thomas-mangin
will look later on - can not find anything obviously wrong
it would like the socket is not setup correctly for MD5 and therefore the kernel is not matching the incoming packet to the socket but AFAICS all was setup fine .
sanjmonkey
@sanjmonkey
ok thomas thanks for checking. enjoy your day!
Thomas Mangin
@thomas-mangin
going to Jitsu - will be mid aft. I have an idea.
Thomas Mangin
@thomas-mangin
There is a connect.conf file with the test.conf ..
using it works …
the connection is not accepted as the source IP is not right but it establishes
could you let me know what is at the other end of the session ?
I will be back later on today - ttfn
sanjmonkey
@sanjmonkey
hi @thomas-mangin. seen your findings. I agree - if it stays local it binds, and capture shows the sending and receiving of TCP MD5. In this case the other end is an identical 14.04 / exa3.4.8. Trying with a Cisco router (IOS 15 something) shows the same, with MD5 on it never makes it to exa.
Thomas Mangin
@thomas-mangin
hum ...
sanjmonkey
@sanjmonkey
I’m going to investigate other OS’ and hypervisors vs physical (in case it is some offloading in vswitch perhaps)
Thomas Mangin
@thomas-mangin
can you try master on the other node .. in case there is an issue..
yes - it could be .. I will try to test between two host later on ( currently working on some other code )
sanjmonkey
@sanjmonkey
other node is 3.4.8 master (not from your tree)
I’ll let you know how it goes
Thomas Mangin
@thomas-mangin
thanks
Thomas Mangin
@thomas-mangin
@sanjmonkey did you find what the problem was - I am waiting to make sure the issue is not with my code before releasing 3.4.9
sanjmonkey
@sanjmonkey
@thomas-mangin nothing conclusive yet. can reproduce in every VM I’ve had access to, though not had time to test with anything physical yet. Have turned off tcp segment offload off in vm, doesn’t appear to make a difference
Thomas Mangin
@thomas-mangin
Thank you for the update
I will try to get it tested in a real machine tomorrow
sanjmonkey
@sanjmonkey
@thomas-mangin I’ve had fun trying some ‘real’ machines today. Surprising how scarce they’ve become. So far I’ve only managed to get my hands on, OSX (message saying no support in Darwin), Arch linux (no TCP_MD5SIG in kernel), Raspbian (currently recompiling kernel to support TCP_MD5SIG, yes, really)….!
have stolen a laptop and installing a current (supported) OS. let you know how I get on shortly
Thomas Mangin
@thomas-mangin
Good luck
My laptop security upgrade seems to have gone wrong but then when I last updated mac os it took hours to do
sanjmonkey
@sanjmonkey
This message was deleted
sanjmonkey
@sanjmonkey
tested ‘real’ fedora 20 machine. same behaviour. exa does not see the incoming md5
Thomas Mangin
@thomas-mangin
That the part where I £%$^^%&%^&£$%&£$%^$% and start to wonder why ...
Thank you very much for checking
I will have a look at BIRD’s code and try to find the difference - that’s the cheating way
Thomas Mangin
@thomas-mangin
with my current workload - it may take a few days.
sanjmonkey
@sanjmonkey
no problem! have you managed to replicate it?
Thomas Mangin
@thomas-mangin
Not had an opportunity (and I changed wholesales provider for DSL and my home line was last night 512k with 5% to 20% packet loss - not ideal to do anything) :-(
The joy of being guinea pig !
Ian bobbitt
@icb-
I'm working on modernizing a project that currently uses zebra (yes...) MRT dumps of the routing table. We're using the routing table snapshots to resynchronize and then replay the updates. Is there a way to get ExaBGP to dump out per-peer route information, or am I going to have to keep track of the entire routing table and write out these snapshots myself? I see there's a "show routes [extensive]" command, but it doesn't seem to do what I want.
Thomas Mangin
@thomas-mangin
Hi @icb- I am unclear on what you want to exactly do …
You will have to write a little bit of code ( because everything in ExaBGP requires a little script to do what you need )
But it should be able to do whatever one would want
Ian bobbitt
@icb-
I know I'll have to write some code. The question is whether I can pass a "show routes received" command (or similar) and get back a dump back of what routes are currently active from a peer, or whether I need to keep track of that myself.
sanjmonkey
@sanjmonkey
@icb- I think you will need to keep track of this yourself, processing the update and withdraw messages
Ian bobbitt
@icb-
That's what I was afraid of. I was just hoping I could get away without it when I saw that exabgp.rib.RIB has both an incoming and outgoing exabgp.rib.store.Store, but the incoming doesn't appear to ever have anything in it.
Thomas Mangin
@thomas-mangin
Correct - we do not store incoming route ATM
You need to create an helper application and handle the routes there