notify_push new version 1.3.3 won't start (Could not automatically determine the process-level CryptoProvider from Rustls crate features.)

[Bad-and-Mad]

I upgraded from version 1.3.1 to 1.3.3. The behavior is the same as in version 1.3.2. The service won't start.
After downgrading to version 1.3.1, everything works again.

Di 19. Mai 07:57:00 CEST 2026
=============================

notify_push new version available: 1.3.3
notify_push updated
end_to_end_encryption new version available: 2.1.1
end_to_end_encryption updated
whiteboard new version available: 1.5.8
whiteboard updated

----------------------------------------------------------------------------------------------------------


systemd[1]: Starting Push daemon for Nextcloud clients...
notify_push[7898]: Error:   × Main thread panicked.
notify_push[7898]:   ├─▶ at /nix/store/0z2r3kypvq12cl33jkaabbifpi8jhd9z-crates-io-dependencies/
notify_push[7898]:   │   rustls-0.23.37-
notify_push[7898]:   │   758025cb5fccfd3bc2fd74708fd4682be41d99e5dff73c377c0646c6012c73a4/src/
notify_push[7898]:   │   crypto/mod.rs:249:14
notify_push[7898]:   ╰─▶
notify_push[7898]:       Could not automatically determine the process-level CryptoProvider from
notify_push[7898]:       Rustls crate features.
notify_push[7898]:       Call CryptoProvider::install_default() before this point to select a
notify_push[7898]:       provider manually, or make sure exactly one of the 'aws-lc-rs' and
notify_push[7898]:       'ring' features is enabled.
notify_push[7898]:       See the documentation of the CryptoProvider type for more information.
notify_push[7898]:
notify_push[7898]:   help: set the `RUST_BACKTRACE=1` environment variable to display a
notify_push[7898]:         backtrace.
systemd[1]: notify_push.service: Main process exited, code=exited, status=101/n/a
systemd[1]: notify_push.service: Failed with result 'exit-code'.
systemd[1]: Failed to start Push daemon for Nextcloud clients.


----------------------------------------------------------------------------------------------------------

sudo -u http RUST_BACKTRACE=1 /big/www/xxx.de/apps/notify_push/bin/x86_64/notify_push /big/www/xxx.de/config/config.php
Error:   × Main thread panicked.
  ├─▶ at /nix/store/0z2r3kypvq12cl33jkaabbifpi8jhd9z-crates-io-dependencies/rustls-0.23.37-
  │   758025cb5fccfd3bc2fd74708fd4682be41d99e5dff73c377c0646c6012c73a4/src/crypto/mod.rs:249:14
  ╰─▶
      Could not automatically determine the process-level CryptoProvider from Rustls crate features.
      Call CryptoProvider::install_default() before this point to select a provider manually, or make sure exactly one of the 'aws-lc-rs' and 'ring'
      features is enabled.
      See the documentation of the CryptoProvider type for more information.

         0: 0x7f8ef5f12f3c - __rustc[9e6a08e89e4b9111]::rust_begin_unwind
         1: 0x7f8ef56c7e0b - core::panicking::panic_fmt::hd8b74df60901cb81
         2: 0x7f8ef56c8123 - core::option::expect_failed::h8ba07abeedb8fcb8
         3: 0x7f8ef5a60887 - futures_util::future::try_future::TryFutureExt::try_poll_unpin::hc92f5062a86ca7b9
         4: 0x7f8ef5a4634f - <futures_util::future::select_ok::SelectOk<Fut> as core::future::future::Future>::poll::hf92fc8b64b208c20
         5: 0x7f8ef5991bed - redis::client::Client::get_multiplexed_async_connection_inner::{{closure}}::h04a6e558e4b888ea
         6: 0x7f8ef59c25f1 - notify_push::App::self_test::{{closure}}::h0eb587c302f77c0b
         7: 0x7f8ef59bb6b5 - notify_push::run::{{closure}}::h6dd22ffc6fbf1069
         8: 0x7f8ef59f5b66 - notify_push::main::h93d4a71856d78811
  help: set the `RUST_BACKTRACE=1` environment variable to display a backtrace.

----------------------------------------------------------------------------------------------------------

[chrissi55]

In my case the error message is slightly different but maybe seems to concern the same cause

░░ A start job for unit notify_push.service has begun execution.
░░ 
░░ The job identifier is 3099.
Mai 19 07:23:24 cloud notify_push[17031]: [2026-05-19 07:23:24.196496 +02:00] WARN [rustls_platform_verifier::verification::others] /nix/store/0z2r3kypvq12cl33jkaabbifpi8jhd9z-crates-io-dependencies/rustls-platform-verifier->
Mai 19 07:23:24 cloud notify_push[17031]: Error:   × Main thread panicked.
Mai 19 07:23:24 cloud notify_push[17031]:   ├─▶ at /nix/store/0z2r3kypvq12cl33jkaabbifpi8jhd9z-crates-io-dependencies/
Mai 19 07:23:24 cloud notify_push[17031]:   │   warp-0.3.7-
Mai 19 07:23:24 cloud notify_push[17031]:   │   4378d202ff965b011c64817db11d5829506d3404edeadb61f190d111da3f231c/src/
Mai 19 07:23:24 cloud notify_push[17031]:   │   server.rs:291:27
Mai 19 07:23:24 cloud notify_push[17031]:   ╰─▶ error binding to [::]:7867: error creating server listener: Address
Mai 19 07:23:24 cloud notify_push[17031]:       family not supported by protocol (os error 97)
Mai 19 07:23:24 cloud notify_push[17031]:   help: set the `RUST_BACKTRACE=1` environment variable to display a
Mai 19 07:23:24 cloud notify_push[17031]:         backtrace.
Mai 19 07:23:24 cloud systemd[1]: notify_push.service: Main process exited, code=exited, status=101/n/a
░░ Subject: Unit process exited
░░ Defined-By: systemd 

In my case i have set my nextcloud server (sysctl.conf) - my NGinx and my Reverse Proxy (also Nginx) fix to use IPv4 only.
so every listener on [::]:Port wont find any "door".

That i presume relies on the changes that have been mentioned on github
#687

My problem is, i don't know any way to set an environment variable or something like that to re-fix notify_push to just listen on IPv4 addresses and to ignore IPv6 (what is enabled by default with 1.3.3)

Any idea of the developers are appreciated.
Thanks in advance.

[Bad-and-Mad]

In my case the error message is slightly different but maybe seems to concern the same cause

░░ A start job for unit notify_push.service has begun execution.
░░ 
░░ The job identifier is 3099.
Mai 19 07:23:24 cloud notify_push[17031]: [2026-05-19 07:23:24.196496 +02:00] WARN [rustls_platform_verifier::verification::others] /nix/store/0z2r3kypvq12cl33jkaabbifpi8jhd9z-crates-io-dependencies/rustls-platform-verifier->
Mai 19 07:23:24 cloud notify_push[17031]: Error:   × Main thread panicked.
Mai 19 07:23:24 cloud notify_push[17031]:   ├─▶ at /nix/store/0z2r3kypvq12cl33jkaabbifpi8jhd9z-crates-io-dependencies/
Mai 19 07:23:24 cloud notify_push[17031]:   │   warp-0.3.7-
Mai 19 07:23:24 cloud notify_push[17031]:   │   4378d202ff965b011c64817db11d5829506d3404edeadb61f190d111da3f231c/src/
Mai 19 07:23:24 cloud notify_push[17031]:   │   server.rs:291:27
Mai 19 07:23:24 cloud notify_push[17031]:   ╰─▶ error binding to [::]:7867: error creating server listener: Address
Mai 19 07:23:24 cloud notify_push[17031]:       family not supported by protocol (os error 97)
Mai 19 07:23:24 cloud notify_push[17031]:   help: set the `RUST_BACKTRACE=1` environment variable to display a
Mai 19 07:23:24 cloud notify_push[17031]:         backtrace.
Mai 19 07:23:24 cloud systemd[1]: notify_push.service: Main process exited, code=exited, status=101/n/a
░░ Subject: Unit process exited
░░ Defined-By: systemd 

In my case i have set my nextcloud server (sysctl.conf) - my NGinx and my Reverse Proxy (also Nginx) fix to use IPv4 only. so every listener on [::]:Port wont find any "door".

That i presume relies on the changes that have been mentioned on github #687

My problem is, i don't know any way to set an environment variable or something like that to re-fix notify_push to just listen on IPv4 addresses and to ignore IPv6 (what is enabled by default with 1.3.3)

Any idea of the developers are appreciated. Thanks in advance.

But that seems to be a different problem.

In my case:

notify_push[7898]:       Could not automatically determine the process-level CryptoProvider from
notify_push[7898]:       Rustls crate features.
notify_push[7898]:       Call CryptoProvider::install_default() before this point to select a
notify_push[7898]:       provider manually, or make sure exactly one of the 'aws-lc-rs' and
notify_push[7898]:       'ring' features is enabled.
notify_push[7898]:       See the documentation of the CryptoProvider type for more information.

In your case:

Mai 19 07:23:24 cloud notify_push[17031]:   ╰─▶ error binding to [::]:7867: error creating server listener: Address
Mai 19 07:23:24 cloud notify_push[17031]:       family not supported by protocol (os error 97)

[chrissi55]

ok just appeded my post here because the header of the issue was/is

notify_push new version 1.3.3 won't start

But i solved my prob by appending "BIND=127.0.0.1" as environment variable to notify_push.service file.

[RomanBAH]

In my case the error message is slightly different but maybe seems to concern the same cause

░░ A start job for unit notify_push.service has begun execution.
░░ 
░░ The job identifier is 3099.
Mai 19 07:23:24 cloud notify_push[17031]: [2026-05-19 07:23:24.196496 +02:00] WARN [rustls_platform_verifier::verification::others] /nix/store/0z2r3kypvq12cl33jkaabbifpi8jhd9z-crates-io-dependencies/rustls-platform-verifier->
Mai 19 07:23:24 cloud notify_push[17031]: Error:   × Main thread panicked.
Mai 19 07:23:24 cloud notify_push[17031]:   ├─▶ at /nix/store/0z2r3kypvq12cl33jkaabbifpi8jhd9z-crates-io-dependencies/
Mai 19 07:23:24 cloud notify_push[17031]:   │   warp-0.3.7-
Mai 19 07:23:24 cloud notify_push[17031]:   │   4378d202ff965b011c64817db11d5829506d3404edeadb61f190d111da3f231c/src/
Mai 19 07:23:24 cloud notify_push[17031]:   │   server.rs:291:27
Mai 19 07:23:24 cloud notify_push[17031]:   ╰─▶ error binding to [::]:7867: error creating server listener: Address
Mai 19 07:23:24 cloud notify_push[17031]:       family not supported by protocol (os error 97)
Mai 19 07:23:24 cloud notify_push[17031]:   help: set the `RUST_BACKTRACE=1` environment variable to display a
Mai 19 07:23:24 cloud notify_push[17031]:         backtrace.
Mai 19 07:23:24 cloud systemd[1]: notify_push.service: Main process exited, code=exited, status=101/n/a
░░ Subject: Unit process exited
░░ Defined-By: systemd 

In my case i have set my nextcloud server (sysctl.conf) - my NGinx and my Reverse Proxy (also Nginx) fix to use IPv4 only. so every listener on [::]:Port wont find any "door".

That i presume relies on the changes that have been mentioned on github #687

My problem is, i don't know any way to set an environment variable or something like that to re-fix notify_push to just listen on IPv4 addresses and to ignore IPv6 (what is enabled by default with 1.3.3)

Any idea of the developers are appreciated. Thanks in advance.

Have same trouble but in nextcloud-aio docker version and cant appending "BIND=127.0.0.1" as environment variable to notify_push.service file.

[chrissi55]

I am not familiar with AiO -> docker for nextcloud / notify_push
I found some info about a file called gai.conf
this file should be located as /etc/gai.conf on the host

In that file should be a line
#precedence ::ffff:0:0/96 100

uncomment as
precedence ::ffff:0:0/96 100

and restart docker - maybe that solves your problem.

sudo systemctl restart docker

docker restart nextcloud-aio-notify-push
docker restart nextcloud-aio-nextcloud

[Bad-and-Mad]

sudo -u http RUST_BACKTRACE=1 /big/www/xxx.de/apps/notify_push/bin/x86_64/notify_push /big/www/xxx.de/config/config.php
Error:   × Main thread panicked.
  ├─▶ at /nix/store/0z2r3kypvq12cl33jkaabbifpi8jhd9z-crates-io-dependencies/rustls-0.23.37-
  │   758025cb5fccfd3bc2fd74708fd4682be41d99e5dff73c377c0646c6012c73a4/src/crypto/mod.rs:249:14
  ╰─▶
      Could not automatically determine the process-level CryptoProvider from Rustls crate features.
      Call CryptoProvider::install_default() before this point to select a provider manually, or make sure exactly one of the 'aws-lc-rs' and 'ring'
      features is enabled.
      See the documentation of the CryptoProvider type for more information.

         0: 0x7f8ef5f12f3c - __rustc[9e6a08e89e4b9111]::rust_begin_unwind
         1: 0x7f8ef56c7e0b - core::panicking::panic_fmt::hd8b74df60901cb81
         2: 0x7f8ef56c8123 - core::option::expect_failed::h8ba07abeedb8fcb8
         3: 0x7f8ef5a60887 - futures_util::future::try_future::TryFutureExt::try_poll_unpin::hc92f5062a86ca7b9
         4: 0x7f8ef5a4634f - <futures_util::future::select_ok::SelectOk<Fut> as core::future::future::Future>::poll::hf92fc8b64b208c20
         5: 0x7f8ef5991bed - redis::client::Client::get_multiplexed_async_connection_inner::{{closure}}::h04a6e558e4b888ea
         6: 0x7f8ef59c25f1 - notify_push::App::self_test::{{closure}}::h0eb587c302f77c0b
         7: 0x7f8ef59bb6b5 - notify_push::run::{{closure}}::h6dd22ffc6fbf1069
         8: 0x7f8ef59f5b66 - notify_push::main::h93d4a71856d78811
  help: set the `RUST_BACKTRACE=1` environment variable to display a backtrace.

The backtrace confirms the problem exactly: The manually executed version crashes immediately upon startup because the binary file is hard-coded to require a decision that it cannot make on its own. This crash affects both versions 1.3.2 and 1.3.3, as the problem is embedded in the source code. The environment variable CRYPTO_PROVIDER=ring is ignored because the application does not even reach the point of querying the variable at that stage (it crashes immediately upon loading the rustls library). To fix the error immediately, you must manually downgrade to the working version 1.3.1.