I received my second probe* yesterday (different ASNs) and it is connects via SixXS for IPv6 as the ISP isn't supplying native v6 (despite repeatedly asking when they will catch up on something around for twenty years!). It booted fine and the appear no issues with the IPv4 connectivity but the v6 is going up and down every few minutes (for around the same period.) Not finding anything relevant after searching the forum (nor on google) I'm trying to discover why this might be. My other machines are not experiencing any difficulty with v6 connectivity and the probe is connected to the same 16-port switch as everything else. Suggestions of what to investigate next very welcome! * https://atlas.ripe.net/probes/28800/ Sent via RIPE Forum -- https://www.ripe.net/participate/mail/forum
On Thu, May 5, 2016 at 3:11 PM, Alison Wheeler <ripecontact@markyate.net> wrote:
... but the v6 is going up and down every few minutes (for around the same period.)
For many months, one of my Atlas probe (v1/v2, firmware 4730) has been running quite nicely through a SixXS IPv6 tunnel. But starting two or three months ago, this probe goes up and down every two to four minutes. As soon as I block all IPv6 connections to/from the probe, it is connectivity over IPv4 only (behind a NAT firewall) is stable and the probe stays up. In another location, a second Atlas probe (v3, firmware 4730) runs stable with dual-stack through a IPv6 tunnel from Hurricane Electric, and with a very similar rule-set on the firewall (pfSense/FreeBSD). Currently, I am looking potential causes: a) My firewall (pf on OpenBSD) rules over-block incoming ICMP6 requests, if Atlas control hosts should now check if the probe has two-way IPv6 connectivity. b) MTU size mismatch somewhere with my SixXS tunnel setup. c) Static vs. dynamic configuration of the probes for IPv6. I will report any progress/findings. Suggestions welcome, thanks!
Hi, On 2016/05/05 15:11 , Alison Wheeler wrote:
I received my second probe* yesterday (different ASNs) and it is connects via SixXS for IPv6 as the ISP isn't supplying native v6 (despite repeatedly asking when they will catch up on something around for twenty years!). It booted fine and the appear no issues with the IPv4 connectivity but the v6 is going up and down every few minutes (for around the same period.)
Not finding anything relevant after searching the forum (nor on google) I'm trying to discover why this might be. My other machines are not experiencing any difficulty with v6 connectivity and the probe is connected to the same 16-port switch as everything else.
Suggestions of what to investigate next very welcome!
It seems that your IPv6 connection has issues. Around the time the ssh connection fails, I also see IPv6 traceroutes with problems. But it is relatively far from the probe, so it is hard to be sure about the source of the problem. One option would be run a traceroute on your probe with relatively high frequency (once a minute) to see if you can spot any pattern in the traceroute output. Philip
Hi Philip, On Tue, May 10, 2016 at 2:56 PM, Philip Homburg <philip.homburg@ripe.net> wrote:
It seems that your IPv6 connection has issues. Around the time the ssh connection fails, I also see IPv6 traceroutes with problems. But it is relatively far from the probe, so it is hard to be sure about the source of the problem.
Thanks for looking into this. I have just noticed that my v3 Probe #16258 now also has a problem with IPv6. But its in a different site with quite a different setup, which is still under construction, and I plan to look after it later today :-) Just to make sure we are talking about the same v1 Probe #2922 which sits at home (on a DOCSIS cable connection, behind a NAT firewall). This is the one which had worked quite stable for months, even years, also through the SixXs tunnel. Something must have changed earlier this year when it began to go down and up every few minutes, most likely due to some (new?) issue with IPv6. Did you test indeed #2922? And from which side did you origin your traceroutes on IPv6, e.g. from this probe, or to this probe from some other location? Thanks, Rolf
On 2016/05/11 10:07 , Rolf Sommerhalder wrote:
Just to make sure we are talking about the same v1 Probe #2922 which sits at home (on a DOCSIS cable connection, behind a NAT firewall). This is the one which had worked quite stable for months, even years, also through the SixXs tunnel. Something must have changed earlier this year when it began to go down and up every few minutes, most likely due to some (new?) issue with IPv6.
It seems that for this probe the tunnel is often down. I.e., traceroute shows the first hop and nothing more. Philip
participants (3)
-
Alison Wheeler -
Philip Homburg -
Rolf Sommerhalder