-----Original Message----- From: Public [mailto:public-bounces@talk.mikrotik.com.au] On Behalf Of Paul Julian Sent: Tuesday, 10 May 2016 9:23 AM To: public@talk.mikrotik.com.au Subject: [MT-AU Public] EOIP Epic Fail
Hi guys, thought I would post this information to see if anybody had experienced the same issue.
We have an RB1100AHx2 in a data centre that connects to a CCR in another data centre, unfortunately we can't do QINQ between the DC's so we run an EOIP tunnel to connect two vlans for a customer through to there, all has been working fine for about 2 years.
The other day, just out of the blue, the customer said their link had gone down, after a number of hours troubleshooting, convinced that it was a problem on site, we discovered that the EOIP tunnel just stopped passing traffic. Nothing had changed on those two routers for months but all of a sudden
just stopped working. The CCR has some other EOIP tunnels on it which were still working fine, but the 1100 was refusing to work.
We checked firewall rules, vlan ID's, tunnel ID's, everything, and there were no issues, the tunnel just refused to pass data.
I tried changing the tunnel ID and it made no difference, tried re-adding
V3 and some V4 ROS variants would sometimes require an ip address or EoIP tunel deleted and re added for no apparent reason. We haven't noticed this behaviour in any of the recent versions though. We tend to try and run a c-tag(s) inside an s-tag where QinQ with all 0x8100 ether type fails across a third party network segment. Not because we have anything against EoIP, but the overhead is so much lower Kind regards Murray Southwell Tasmanet Pty Ltd Manager - Network Services murray@tasmanet.com.au 0361652511 0418145979 -----Original Message----- From: Public [mailto:public-bounces@talk.mikrotik.com.au] On Behalf Of Mike Everest Sent: Tuesday, 10 May 2016 9:52 AM To: 'MikroTik Australia Public List' Subject: Re: [MT-AU Public] EOIP Epic Fail I've seen something similar several years ago where IP address would simply stop working: an ordinary address on an interface would become unreachable, even from the localhost. The only way to recover would be to remove the address, then re-add. That was way back in some early v3 routerOS I think, and after upgrades, I just forgot that it ever happened - until now ;) So I guess I'm leaning toward software bug - what routerOS version is it? Apart from waiting for it to happen again and try to get some quick diagnostics knocked out (at least grap some pcap data) there's probably not much more to do than wait - and seek feedback from others, of course, for which I'm also interested to hear about! Cheers! Mike. they the
vlans to it, nothing, disabled and enabled it, nothing made a difference.
Out of desperation we created a new EOIP tunnel on that router with a new ID and pointed it to the existing EOIP interface at the other end and it came up straight away.
So apart from a big WTF, has anybody experienced this before ?
Regards Paul _______________________________________________ Public mailing list Public@talk.mikrotik.com.au http://talk.mikrotik.com.au/mailman/listinfo/public_talk.mikrotik.com. au
_______________________________________________ Public mailing list Public@talk.mikrotik.com.au http://talk.mikrotik.com.au/mailman/listinfo/public_talk.mikrotik.com.au