You are here

Tunnel connections dropping in and out

16 posts / 0 new
Last post
wa2ise
wa2ise's picture
Tunnel connections dropping in and out

I've noticed that some but not all my tunnel connections will go dead every so often, but will come back up a few hours later.  I off loaded my "advertised services" web pages out of my nodes, and onto separate OpenWRT routers, as people recommended.  My ISP is Optimum cable.  I suppose some of it might be the tunnel server owners doing maintenance, taking it off-line to fix or upgrade their stuff.  But it seems to happen a little too often for that...  Maybe twice a week.  

KG6JEI
Really hard for any real

Really hard for any real reason to be stated. Best to work with the other end
of the tunnels to see what they are doing. It is however not uncommon for those who utilize tunnels to change around frequently as they are usually in some of the initial stages of learning while they try and get a true RF link in place.

Kd6mtu
drops / olsr

I am seeing the same thing from my end.   I notice I will see a few OLSR restarts and then finally the tunnel to you will go away.... Sometimes a reboot on my end will bring the tunnel back up but not sure if thats just a timing parallel ...

WU2S
WU2S's picture
Tunnel connections

I have been doing a lot of tunnel testing lately - just a normal part of my role in AREDN. The connections that I supply are intended to be a test environment - useful for people who want to see the wider world of a mesh network with a lot of nodes and services. There are times when the connections go up and down as tests are performed or new software is installed. This may be what you are experiencing.

KD1HA
KD1HA's picture
Node drops off the status page but active in the tunnel !?!

I have a client node that drops off the status page but shows active in the tunnel, this also has happened with another server from the same client. All are Ubiquiti nodes. We have refreshed the firmware with the 3.17.1.0RC1 on all 3.

AE6XE
AE6XE's picture
Ref: submitted defect:  http:

Ref: submitted defect:  [LINK REMOVED]

Is this also the case for you?:
1) The scenario seems to occur when there are no RF links to the mesh node with the tunnel(s).    
2) degraded performance--takes 30sec or longer to show node status on the mesh node with the tunnels after a a day or so after boot.

If 'yes', then it is possible the root cause of your issue is the same as this defect.   We are currently building a debug version of the firmware to squish this bug.

Joe AE6XE

KD1HA
KD1HA's picture
Node drops off the status page but active in the tunnel !?!

In a word "yes"! It may help to know that the previous server does not and did not have any RF links as well and is connected to me now as a client (server is clean with no clients) and we did re-load of the firmware to confirm a clean install (to both clients) and I/we can see him fine but the one in question. I will stand by for a resolve of Ticket/218
Thank you,
Denis
 

wa2ise
wa2ise's picture
Yes, that sounds like it.  No

Yes, that sounds like it.  No RF links, and degraded performance, especially when doing a mesh status.  And node status taking a long time too.  To clear it, I have to pull the power supply's power plug (most convenient method for my setup to hard power cycle it).  I have two tunnel clients active on it.  Don't know if that makes the problem worse or not... 

k1ky
k1ky's picture
Tunnel reliability

Several issues here.  If you are on 3.17RC1, you can "revert" back to 3.16 Production and see if it improves. One of the bigger issues here is with "marginal" internet connections.  If there are any interruptions in your connection, the node may drop the tunnel connection and go into an "OLSR Not Running" condition.  All you need to do to clear this is to go to your tunnel client, uncheck the tunnel box (to the node that isn't working - note that the Blue Cloud may still be lit even though the connection has failed) save the setting, wait for confirmation - then re-check the tunnel client box and save again.  Once the settings have been saved, your olsr should be running again and the connection restored... until the next connection drop.
There is no need to reboot the node.
 

KD1HA
KD1HA's picture
The link above does not

The link above does not connect to ticket 218 [LINK REMOVED] but if what I can fined for 218 it appears to be an AirRouter issue however in this case it's a NanoStation M2 at a school in Middletown, Rhode Island N1ASA mounted on the roof with no RF connections as yet. All the setup information looks good, SSID and so on and I realize that it is the same software and may be, in part the same issue but I just wanted to clarify that. If you need any other information please let me know.

Denis KD1HA      

wa2ise
wa2ise's picture
I noticed that my Aredn nodes

I noticed that my Aredn nodes have had no OLSR restarts for the past 2 weeks now.  I haven't done anything here, so far as I know.  FWIW...

K5DLQ
K5DLQ's picture
Are you tunneled to someone?

Are you tunneled to someone?
if so, who?

wa2ise
wa2ise's picture
My Airrouter is tunneled to

My Airrouter is tunneled to kd6mtu.  The Airrouter has been up 25 days, no OLSR restarts.  However, my Bullet's tunnel connections were not connected.  I still had lots of mesh status activity, but that must have been thru the Airrouter's tunnel to kd6mtu.  I rebooted the Bullet, and the tunnel to w6bb came back up.  But my tunnel to WU2S is still dead. 

So my above observation is probably not valid...  Having just one tunnel client connection seems to work fine, but having two seems to lead to OLSR restarts?

WU2S
WU2S's picture
WU2S had several outages

I had about 4 outages a few days ago. Two were due to PSE&G power failures and two were due to a Verizon FiOS upgrade.One outage of each type lasted for about 90 minutes.

wa2ise
wa2ise's picture
Just checked, the tunnel to

Just checked, the tunnel to you is working now.  I'll see if, with 2 tunnel client connections running, if I get any OLSR restarts on my Bullet in the next week or so...

Pubic Service Electric and Gas is usually very reliable at my QTH.  Turns out the guy who was in charge of distribution for PSEG in Bergen County lived just a few doors from me. smiley  There's a hospital about a mile away from here, and hospitals are given top priority for reliable service, and the local substation is about 2 miles away in roughly the other direction.  So my power may be fed off that hospital link..  We didn't lose power during that massive blackout in 2003 (though the line voltage did oscillate gradually up to around 135V and down to around 90V every few minutes every five or so minutes). 

Way back in 1965 we did not lose power in that big Northeast blackout.  Though most NYC TV stations were dead. 

kj6dzb
kj6dzb's picture
W6BB checking in!

W6BB checking in!

I do run the nodes on our campus network not your standard isp. Is the isp shaping traffic?

I did have some trubble running 2 nodes w/ 2 tunnel servers off one wan port. I thought the vlan switch was the problem and dropping packets. So i moved the server to only one node and things stabilized. I have 5 clients, with no olsrd restarts.

I know i have funky power at the site. But that has never effected the node stability.

I do keep one rf node connected. With that said I would suspect that olsrd may be timing out because of how some isp's treat packet traffic and or vlan switch reliability.

73 mathison kj6dzb

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer