You are here

Tunnel connection isn't fault tolerant

6 posts / 0 new
Last post
AH6LE
Tunnel connection isn't fault tolerant

Not sure if this belongs here or not but I noticed something.

I am running a hAP Lite that tunnels into another hAP Lite on our local mesh. Mine also is connected to my LAN's router for obvious internet access for the tunnel. My Internet is provided by local cable.

Yesterday, we experienced a short (10 minutes or so) "outage" where I lost all internet connectivity, but never power. When internet service started working again, I didn't give it a 2nd thought until later in the day when I wanted to do something on a remote node. As it turns out, I was no longer connected to the tunnel server and it took a reboot of my hAP in order to re-establish the tunnel.

Seems to me the tunnel client/server connection should be more fault tolerant than that, no?
 

K6CCC
K6CCC's picture
I have never had a problem

I have never had a problem with either of my hAP nodes failing to reconnect after an internet outage without having to power cycle them.  My primary hAP-at-home normally has several tunnels where the hAP is the client, and several tunnels where the hAP is the server.  In both conditions, the tunnels reconnect.

 

AH6LE
Maybe it's the version FW I'm

Maybe it's the version FW I'm running?

3.20.3.1

K6CCC
K6CCC's picture
I doubt it's the version.

I doubt it's the version.  I have run every release version since 3.19.something and damn near every nightly build in between.
 

AH6LE
Not sure what to say then. As

Not sure what to say then. As a test, I disconnected the cable from my cable modem, waited 5 minutes and reconnected it (didn't power down anything). Internet service came back within a minute but no tunnel connection. I had to reboot the hAP again in order to re-establish.

Weird......

K6CCC
K6CCC's picture
Worked OK for me.

I just did a test with my hAP-at-Home node.  Prior to the test, there were three tunnels where my hAP was the tunnel server, and two tunnels where my hAP was the tunnel client.  For the test, I disabled VLAN 1 in my router.  This would result in all packets from the internet destined to the tunnel, and all packets from the hAP outbound on VLAN 1 ending up in the bit bucket.  The physical port 1 on the hAP would still be connected to the LAN switch, so there would be no change of status of the port itself.  It took a couple minutes before the hAP showed that all the tunnels were down, but as expected, they did go down.

After a couple minutes, I re-enabled VLAN 1 in my router and started watching the tunnel status pages in the hAP.  At no point did I check or un-ckeck the "Enabled" checkboxes, nor "Save Changes" - just toggled between the tunnel client and tunnel server pages.  The first tunnel where I am the server came up in about 30 second, and all except one client connection came up within two minutes.  I waited eight minutes for the one client connection, and it would not come up even after I cycled the "Enabled" checkbox (saving changed each time), and I could not ping the DNS entry for it, so I am suspecting something on the other end.

At the moment, that hAP is running nightly build 300.

 

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer