At 16:58 3/27/2016 +0000, nusenu wrote:

Hi,

...

For the past several days the number of relays measured by my bwuath (whose data is given my maatuska) have been slowly declining. Originally I had an alert when it was below 8800, a few days I noticed it going below that. By today it's decreased to below 8700 to 8664.

It's not the lowest bwauth measuring, but I don't see anything obviously wrong with my bwauth to indicate the decline. (The number I mentioned is number of lines in the bwscan.V3BandwidthsFile file, not the measured relays that make it into the consensus - but the comparison I just made did use the consensus count.)

If "past several days" matches with "since 2016-03-24", then I would not be surprised if the Global Frag Networks group [1] caused this - at least partially.

Many of the unmeasured relays are from the global frag group (since they just joined the network)

https://github.com/ornetstats/stats#stats-on-measured-and-recommended-flag

Many new relays means a rise in unmeasured relays and if at the same time a few relays leave then that would explain the decline even for absolute values.

[1] http://article.gmane.org/gmane.network.onion-routing.ornetradar/1135 http://article.gmane.org/gmane.network.onion-routing.ornetradar/1132

Interesting. Perhaps the arrival of bunches of new relays is a trigger for or exacerbates the bug where BWauths skip measurements, scanner/band #1 esp.