this is dual-stack by default on linux, resolves
issues with nginx using `localhost` and randomly
choosing between 127.0.0.1 and [::1], causing
intermittent upstream issues
Signed-off-by: strawberry <strawberry@puppygock.gay>
The original implementation of this was really weird, so I restructed it
a lot while debugging, and am just gonna leave the restructured version.
Root cause of the segfault seems to be that upstream nixpkgs liburing
derivation is generating both static and dynamic libraries, causing
rocksdb to statically link liburing in a dynamic build, pulling in some
allocator stuff at the same time. I created a PR[1] to fix this upstream,
but it probably won't be available on nixos-unstable for quite a while,
so we can also patch it locally.
[1]: https://github.com/NixOS/nixpkgs/pull/314945
I talked to somebody yesterday in #conduwuit:puppygock.gay that was using
this output in their system config. The dynamically-linked jemalloc build
is quite fragile, and is not tested by anything else in CI. We want to
make sure we don't break it again in the future.
This failed to inherit the fix from bec507d739
because the crane package's buildInputs become propagatedBuildInputs in
a static stdenv, but become normal buildInputs in a dynamic stdenv. Since
we were only pulling propagatedBuildInputs into the devshell, dynamically
linked devshells did not include the rust-jemalloc-sys package. This
causes tikv-jemalloc-sys to build it's own static jemalloc package, and
we end up loading libc before jemalloc at runtime.
It turns out that this was actually fixed by
bec507d739 and
857ac42aac, but we didn't identify it at the
time. Notably, the `dynamic` devshell is still broken.
needed for hot reloading but rpath being true by default
causes linker errors on lld because of the sad rpath bug
Signed-off-by: strawberry <strawberry@puppygock.gay>
for a very long time, if a remote server responded to us with
a valid but unsuccessful (HTTP 4xx) response and the caller was the
`send_federation_request` function, we may find ourselves
with a warning message only containing the destination's
server name which was very unhelpful. the true error was
buried away in trace logs. this would primarily be noticed
with server key fetch requests from us.
conduit has been throwing away the ruma request error: https://gitlab.com/famedly/conduit/-/blame/next/src/utils/error.rs#L62
before: 2024-05-23T04:45:02.930224Z WARN router:{path=/_matrix/client/v3/publicRooms}:handle: conduit_api::client_server::directory: Failed to return our /publicRooms: matrix.org
after: 2024-05-23T05:05:02.435272Z WARN router:{path=/_matrix/client/v3/publicRooms}:handle: conduit_api::client_server::directory: Failed to return our /publicRooms: matrix.org: [401 / M_UNAUTHORIZED] Failed to find any key to satisfy: _FetchKeyRequest(server_name='your.server.name', minimum_valid_until_ts=1716440702337, key_ids=['ed25519:RQB3XPQX'])
Signed-off-by: strawberry <strawberry@puppygock.gay>
Jason Volk