excurso authored Sep 26, 2024 and David Goulet committed Oct 10, 2024

Resources allocated by cpuworker weren't being freed on clean shutdown.
This applies for worker threads, worker thread pool, reply queue, reply
event, ...

Conflux circuit building was ignoring the "predicted ports" feature,
which aims to make Tor stop building circuits if there have been
no user requests lately. This bug led to every idle Tor on the
network building and discarding circuits every 30 seconds, which
added overall load to the network, used bandwidth and battery from
clients that weren't actively using their Tor, and kept sockets open
on guards which added connection padding essentially forever.

Bug went in on commit 39c2927d when we added preemptive conflux circuit
pools.

Fixes bug 40981; bugfix on 0.4.8.1-alpha.

We're about to add a few more steps, which would make the single-line
format a bit unwieldy.

excurso authored Oct 30, 2024 and David Goulet committed Oct 31, 2024

See issue #40989

Jim Newsome authored Nov 18, 2024 and David Goulet committed Nov 19, 2024

* buster is now very old
* all the other jobs run bullseye
* This will facilitate switching to a TPA-maintained image (there is no
  buster image).

Jim Newsome authored Nov 18, 2024 and David Goulet committed Nov 19, 2024

This is to work around
https://github.com/docker-library/official-images/issues/17896

Previously we would incorrectly detect that ipv6 isn't supported if the
ping binary isn't present (as it may not be in a relatively stripped
down container image), or if ICMP packets aren't permitted (as they
often aren't in containers).

We don't really have a need to run these network tests on non-IPv6
systems, so it makes more sense to just run them unconditionally.

This should be a pure refactor to simplify the logic a bit now that we
no longer probe for IPv6 support.