[ixpmanager] New install step by step
Vladislav Bidikov
vladislav.bidikov at finki.ukim.mk
Mon Jul 23 17:36:22 IST 2018
Hi Nick,
I will try to comment inline 😃
Bare with me as my email client seems a bit with his own mind most of the time ...
Thanks for the super fast response...
--
V.B
FCC
________________________________
From: Nick Hilliard <nick at inex.ie>
Sent: Monday, July 23, 2018 6:19 PM
To: INEX IXP Manager Users Mailing List; Vladislav Bidikov
Subject: Re: [ixpmanager] New install step by step
Vladislav Bidikov wrote on 23/07/2018 16:51:
>> I chose Customer type as Route Server (although examples show pro-bono
>> as option) is this ok?
>yep, that's correct.
>> Now several questions emerge:
>>
>> 1.
>>
>> I would like to add the second Route Server (separate VM) - can i just
>> add it under routes the same as i did for RS1 ?
>yes.
Ok. this is settled...
>> 2.
>>
>> Can i add another customer (with private AS number) for Route Colector -
>> and than go from there ?
>You should set the Route Collector up as being owned by the IXP itself.
>It would be a good idea for the IXP to have its own ASN.
See this screen shot:
http://prntscr.com/k9ztzm
Since i only got one AS number (for the peering/RS) maybe we made a mistake ...
Can i use the same As also for Route collector - wince the services of the IXP are not in this ip/AS space ?
>> 3.
>>
>> Since all these VMs (2 x RS, 1 x Rc, 1 x AS112) will be connected on the
>> same Port (the Eth-Trunk) how will IXP manager handle this since there
>> will be 3 customers ?
>The VM acts as a virtual switch. In fact, this usually causes
>monitoring problems because most (all?) VM snmp implementations don't
>support per-interface monitoring.
>
>The physical switch port configuration can be done manually.
This trunk to the VM based on Debian (which run on VmWare) will be working fine since the port itself is as Eth-Trunk and MRTG is monitoring it that way...
Also, after we resolve (4) with a separate port for FCSE, the only traffic there will be for RS/RC/AS112 and the 2x1G trunk is more for redundancy purposes...
Config of switches will be done completely manually...
>> 4.
>>
>> Since i expect one real customer (ourselves as FCSE - with AS 52188)
>> also will come connected on the same trunk (to save few ETH ports) how
>> to handle this?
>um, you don't really want to do this. Each real customer should have
>their own physical port with appropriate port security mechanisms. FCSE
>might run the IXP but it's not the IXP, and each should have their own port.
Ok, new 1G (copper/optic) combo port for this ...
>> 5.
>>
>> Also, i enabled MRTG grapher which i see all files are generated/updated
>> in /srv/mrtg but the web page shows imaged not found:
>>
>> http://prntscr.com/k9zd66
>>
>> and, view source does not help since i cannot debug where does the img
>> is loaded from...
>These files should come straight from /srv/mrtg. Have you configured
>the following variables in .env:
>
>GRAPHER_BACKEND_MRTG_WORKDIR="/srv/mrtg"
>GRAPHER_BACKEND_MRTG_LOGDIR="/srv/mrtg"
These where there, after i got your reply and hit refresh graphs started showing without intervention ....
So this is resolved...
Best regards,
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.inex.ie/pipermail/ixpmanager/attachments/20180723/5e8c2b91/attachment-0001.html>
More information about the ixpmanager
mailing list