[aprssig] DNS for APRSIS Servers

[Tom Hayward]

On Thu, Feb 19, 2015 at 5:33 AM, Jason KG4WSV <kg4wsv at gmail.com> wrote:
> On Wed, Feb 18, 2015 at 9:58 PM, Max Harper <kg4pid at yahoo.com> wrote:
> > Is someone messing around with the DNS entries for the aprsis server
> > T2TEXAS?
> > I ask because my igate only knows about one server "texas.aprs2.net"  but I
> > am currently connected to "iad.aprs2.net".
>
> Don't know, but juggling DNS records is a method of dealing with
> outages, so there may be some very good sysadmin type reasons that
> you're seeing this.

Jason is right. The Tier2 DNS entries are controlled by an automated
monitoring system. If a problem is detected with a server, the DNS
entry will be pointed at rotate.aprs2.net. The servers in
rotate.aprs2.net are also chosen automatically based on eligibility
criteria.

Off the top of my head, these reasons could cause a server to be
removed from DNS:
- Status page (:14501) not reachable
- Filter port 14580 not acknowledging connections
- No route to host (layer 3)
- Too many users, no open slots
- No uplink connection
- Server sysop marked their server "out of service" for maintenance in
the portal.

Assuming all the previous tests pass, these are the additional scoring
criteria for inclusion in rotate.aprs2.net:
- Round trip time (ping) to server
- Uplink has been stable for more than 15 minutes
- No unplanned downtime in the last 3 days
- Number of available connection slots

These tests are done every five minutes from three
geographically-diverse locations.

You can see the results of the tests here:
http://status.aprs2.net/

Red rows are server errors.
Blue rows are servers included in rotate.aprs2.net.
Blue server names are included in a regional rotate, like noam.aprs2.net.
Yellow indicates a server is down for planned maintenance.

This system is all open source, written by Hessu:
https://github.com/hessu/aprs2net-backend

Tom KD7LXL