The list of TLS versions was added by
8bd2b6eb13 when hackney version was
pinned to 1.15.2. Later hackney version was upgraded
(166455c884) but the list of TLS
versions wasn't removed. From the hackney point of view, this list has
been replaced by the OTP defaults since 0.16.0
(734694ea4e24f267864c459a2f050e943adc6694).
It looks like the same issue already occurred before:
0cb7b0ea84.
A way to test this issue (where example.com is an ActivityPub site
which uses TLSv1.3 only):
$ PLEROMA_CONFIG_PATH=/path/to/config.exs pleroma start_iex
Erlang/OTP 22 [erts-10.7.2.16] [source] [64-bit] [smp:2:2] [ds:2:2:10] [async-threads:1] [hipe]
Erlang/OTP 22 [erts-10.7.2.16] [source] [64-bit] [smp:2:2] [ds:2:2:10] [async-threads:1] [hipe]
Interactive Elixir (1.10.4) - press Ctrl+C to exit (type h() ENTER for help)
iex(pleroma@127.0.0.1)2> Pleroma.Object.Fetcher.fetch_and_contain_remote_object_from_id("https://example.com/@/Nick/")
{:error,
{:tls_alert,
{:protocol_version,
'TLS client: In state hello received SERVER ALERT: Fatal - Protocol Version\n'}}}
With this patch, the output is the expected one:
iex(pleroma@127.0.0.1)3> Pleroma.Object.Fetcher.fetch_and_contain_remote_object_from_id("https://example.com/@/Nick/")
{:error,
{:ok,
%{
"@context" => [
"https://www.w3.org/ns/activitystreams",
"https://w3id.org/security/v1",
%{
"Emoji" => "toot:Emoji",
"Hashtag" => "as:Hashtag",
"atomUri" => "ostatus:atomUri",
"conversation" => "ostatus:conversation",
"featured" => "toot:featured",
"focalPoint" => %{"@container" => "@list", "@id" => "toot:focalPoint"},
"inReplyToAtomUri" => "ostatus:inReplyToAtomUri",
"manuallyApprovesFollowers" => "as:manuallyApprovesFollowers",
"movedTo" => "as:movedTo",
"ostatus" => "http://ostatus.org#",
"sensitive" => "as:sensitive",
"toot" => "http://joinmastodon.org/ns#"
}
],
"endpoints" => %{"sharedInbox" => "https://example.com/inbox"},
"followers" => "https://example.com/@/Nick/followers",
"following" => nil,
"icon" => %{
"type" => "Image",
"url" => "https://example.com/static/media/[...].png"
},
"id" => "https://example.com/@/Nick/",
"inbox" => "https://example.com/@/Nick/inbox",
"liked" => nil,
"name" => "Nick",
"outbox" => "https://example.com/@/Nick/outbox",
"preferredUsername" => "Nick",
"publicKey" => %{
"id" => "https://example.com/@/Nick/#main-key",
"owner" => "https://example.com/@/Nick/",
"publicKeyPem" => "[...]
},
"summary" => "",
"type" => "Person",
"url" => "https://example.com/@/Nick/"
}}
A way to test the reverse proxy bits of this issue (where example.com allows TLSv1.3 only):
iex(pleroma@127.0.0.1)1> Pleroma.ReverseProxy.Client.Hackney.request("GET", "https://example.com", [], [])
{:error,
{:tls_alert,
{:protocol_version,
'TLS client: In state hello received SERVER ALERT: Fatal - Protocol Version\n'}}}
body
- Modify `close/1` function to do the same thing it does for hackney,
which is - close the client rather than the whole connection
- Release the connection when there is no body to chunk
This patch refactors gun pooling to use Elixir process registry and
simplifies adapter option insertion.
Having the pool use process registry instead of a GenServer has a number of advantages:
- Simpler code: the initial implementation adds about half the lines of code it deletes
- Concurrency: unlike a GenServer, ETS-based registry can handle multiple checkout/checkin
requests at the same time
- Precise and easy idle connection clousure: current proposal for closing idle connections in
the GenServer-based pool needs to filter through all connections once a minute and compare their
last active time with closing time. With Elixir process registry this can be done
by just using `Process.send_after`/`Process.cancel_timer` in the worker process.
- Lower memory footprint: In my tests `gun-memory-leak` branch uses about 290mb on peak load (250 connections)
and 235mb on idle (5-10 connections). Registry-based pool uses 210mb on idle and 240mb on peak load
The current failure tracking mechanism will never request anything that
didn't respond with a success, 403, 404, or 5xx codes. This is causing
issues when using in real fediverse because of weird status codes
some software has and timeouts being frequent. This patch changes
failure tracking mechanism to only never request the url again if it
responded with 400, 204, or the body is too large, otherwise it can be
re-requested in 60 seconds.
lain
![https://example.com/static/media/[...].png](https://example.com/static/media/[...].png){kind=link}