You are here

Home » Forums » Regional » California » South Cal counties

So Cal Mesh Status Slow???

5 posts / 0 new
Log in or register to post comments
Last post
Thu, 02/11/2021 - 20:26
#1
km6zpo
km6zpo's picture
So Cal Mesh Status Slow???

Is it just me, or is everybody in the So Cal area experiencing severe slowness (and sometimes nothing at all) when checking their mesh status page on their nodes?  This is my main home node:

http://10.177.158.225/cgi-bin/status
 

KM6ZPO-HOME-WIFI.local.mesh


I have one client tunnel to K6CCC (Glendora - my main tunnel to the Southern California Mesh) and one to a guy in Maryland.  I tried turning off the Maryland tunnel which I host as a tunnel server.  It made no impact.  

My current home speed test is 20 Mbps down, 9 Mbps up.  It's usually 50 Mbs down but again that really shouldn't matter for the crawl I'm seeing.

---mark, KM6ZPO

 

Top
Thu, 02/11/2021 - 21:33
K6CCC
K6CCC's picture
Keep the Maryland tunnel off.
Keep the Maryland tunnel off.  Yes, we have had a lot of issues recently with routing loops messing up OLSR (simple answer).
 
Top
Fri, 02/12/2021 - 10:39
(Reply to #2)
km6zpo
km6zpo's picture
What causes loops?
What would cause loops to happen?  Anyway to prevent that?
Top
Fri, 02/12/2021 - 10:55
K6CCC
K6CCC's picture
Simple.  A connection (RF,
Simple.  A connection (RF, DtD, or tunnel) from node A to Node B, to Node C, to node D, to Node A.  OLSR does not have a good way to handle that when the routing cost (ETX) is similar.  Or worse, one is changing slightly on a regular basis.  For example using the four node loop above, if all four links have an ETX of 1.25, but the link from B to C is changing between 1.20 and 1.30.  In that case if a packet is trying to go from A to C, at times that the B to C ETX is 1.20, then the router A > B > C is better with a combined ETX of 2.45 vs A > D > C which is 2.50.  Five minutes later when the B to C ETX is now 1.30, now the route A > B > C is now 2.55 vs A > D > C is still 2.50 - so A > D > C is better.  A Few minutes later, it changes again.   This is referred to a route flapping and is not good.

The only solution (at this time) is preventing routing loops.  Think hub and spoke mentality.
 
Top
Fri, 02/12/2021 - 11:39
(Reply to #4)
km6zpo
km6zpo's picture
Simple???
Simple.  Hmmm.  Yeah that went way over my head.   I was thinking more along the lines of best practices for tunnel server / client entries to prevent loops.  Or for that matter, best pratices on linking out of area. 

I can understand that keeping a link open to the other side of the world might add some lag, especially if that other side has hundreds of nodes connected to it on the other side.  That make sense.

I have also heard of "global" nodes. Maybe there are even "regional nodes" (i.e. United States, California, West Coast, etc.)  Where are those?  

 
Top

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer