Bismark

Description

This consists of two different packages: bismark-mgmt, which talks to the server and sets up remote tunnels (apart from other things), and bismark-active, which conducts active measurements.

In bismark-mgmt, a probe runs every minute. This probe consists of a single UDP packet to the probe server, and serves two purposes; one as a heartbeat, so that the server knows it’s alive, and also to punch a hole in the NAT (if one exists), so that a reverse-ssh tunnel can be established on-demand from the server.

The bismark-active script runs every 5 minutes. It conducts a subset of the following tests: ping to one of a set of servers, ping to the last-mile router, traceroute to the server, dns lookup time to a set of servers, netperf upstream and downstream tests, ping to the last mile router at the same time as the netperf tests, shaperprobe, and ditg for measuring packet loss.

The output from the test is stored as an xml file in /tmp/bismark/ and, upon completion of test, is immediately uploaded to the measurement server and deleted upon upload.

Testing

bismark-mgmt should report to bdm after first boot, and every minute thereafter, as long as it’s connected to the Internet. First probe takes a few minutes after boot, perhaps up to 5 minutes. Once it reports, we should be able to establish an ssh tunnel to it from the server.

bismark-active is tested by waiting for the data file on the measurement server. The first file should contain all of the above measurement tests; and should have sane values for them.