On Sat, Oct 12, 2019 at 08:04:00AM +0000, nusenu wrote:

would you mind publishing the actual list of fingerprints you rejected this time (2019-10-08) like you did the last time during the removal of heartbleed relays? [1]

Good idea.

I've attached the list of 687 obsolete fingerprints that we rejected on Oct 8, as well as the 8 fingerprints that we've un-rejected since then.

There were a set of 70 relays running 0.3.4.x and with ContactInfo set, which we decided to give a bit more time to upgrade since I had only mailed a few of them recently. We will probably bump them out at some point.

And note that moria1 never applied these reject lines. I instead just directly applied the patch to reject relays running obsolete versions.

So if any of these 687 rejected relays upgraded, their new descriptors will still have been published on, voted about, etc by moria1 the whole time: https://collector.torproject.org/recent/relay-descriptors/ https://collector.torproject.org/recent/relay-descriptors/votes/

In terms of impact to the network, the lost capacity is definitely noticeable, but the actual load on the Tor network hasn't dropped much: https://metrics.torproject.org/bandwidth.html

We can watch the onionperf graphs to see if performance has changed: https://metrics.torproject.org/torperf.html (not much impact yet, maybe that means not much impact total)

Hope this helps, --Roger