[pbs-devel] [RFC PATCH proxmox-backup] tools/async_io: do not error on Accept for StaticIncoming

[pbs-devel] [RFC PATCH proxmox-backup] tools/async_io: do not error on Accept for StaticIncoming

[Dominik Csapak]

in proxmox-backup-proxy, we log and discard any errors on 'accept',
so that we can continue to server requests

in proxmox-backup-api, we just have the StaticIncoming that accepts,
which will forward any errors from the underlying TcpListener

this patch also logs and discards the errors, like in the proxy.
Otherwise it could happen that if the api-daemon has more files open
than the proxy, it will shut itself down because of a
'too many open files' error if there are many open connections

(the service should also restart on exit i think, but this is
a separate issue)

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
---
two remaining issues/questions:

1. maybe we want an accept loop in proxmox-backup-api akin to the one
   in the proxy? this way we have better locality of the error logging,
   more control of e.g. how many connections we accept, etc.

2. is it wise to log every failed connection attempt? i can very easily
   generate massive logspam with a single line of 'ab'. shouldn't we
   e.g. only log once a second how many connections were not accepted?
   this is for the api as well as the proxy

 src/tools/async_io.rs | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/src/tools/async_io.rs b/src/tools/async_io.rs
index 997c02fa..844afaa9 100644
--- a/src/tools/async_io.rs
+++ b/src/tools/async_io.rs
@@ -6,6 +6,7 @@ use std::pin::Pin;
 use std::task::{Context, Poll};
 
 use futures::stream::{Stream, TryStream};
+use futures::ready;
 use tokio::io::{AsyncRead, AsyncWrite, ReadBuf};
 use tokio::net::TcpListener;
 use hyper::client::connect::Connection;
@@ -108,10 +109,15 @@ impl hyper::server::accept::Accept for StaticIncoming {
         self: Pin<&mut Self>,
         cx: &mut Context,
     ) -> Poll<Option<Result<Self::Conn, Self::Error>>> {
-        match self.get_mut().0.poll_accept(cx) {
-            Poll::Pending => Poll::Pending,
-            Poll::Ready(Ok((conn, _addr))) => Poll::Ready(Some(Ok(conn))),
-            Poll::Ready(Err(err)) => Poll::Ready(Some(Err(err))),
+        let this = self.get_mut();
+        loop {
+            match ready!(this.0.poll_accept(cx)) {
+                Ok((conn, _addr)) => return Poll::Ready(Some(Ok(conn))),
+                Err(err) => {
+                    eprintln!("error accepting connection: {}", err);
+                    continue;
+                }
+            }
         }
     }
 }
-- 
2.20.1

[pbs-devel] applied: [RFC PATCH proxmox-backup] tools/async_io: do not error on Accept for StaticIncoming

[Wolfgang Bumiller]

applied

On Thu, Apr 08, 2021 at 02:28:20PM +0200, Dominik Csapak wrote:
> in proxmox-backup-proxy, we log and discard any errors on 'accept',
> so that we can continue to server requests
> 
> in proxmox-backup-api, we just have the StaticIncoming that accepts,
> which will forward any errors from the underlying TcpListener
> 
> this patch also logs and discards the errors, like in the proxy.
> Otherwise it could happen that if the api-daemon has more files open
> than the proxy, it will shut itself down because of a
> 'too many open files' error if there are many open connections
> 
> (the service should also restart on exit i think, but this is
> a separate issue)
> 
> Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
> ---
> two remaining issues/questions:
> 
> 1. maybe we want an accept loop in proxmox-backup-api akin to the one
>    in the proxy? this way we have better locality of the error logging,
>    more control of e.g. how many connections we accept, etc.
> 
> 2. is it wise to log every failed connection attempt? i can very easily
>    generate massive logspam with a single line of 'ab'. shouldn't we
>    e.g. only log once a second how many connections were not accepted?
>    this is for the api as well as the proxy
> 
>  src/tools/async_io.rs | 14 ++++++++++----
>  1 file changed, 10 insertions(+), 4 deletions(-)
> 
> diff --git a/src/tools/async_io.rs b/src/tools/async_io.rs
> index 997c02fa..844afaa9 100644
> --- a/src/tools/async_io.rs
> +++ b/src/tools/async_io.rs
> @@ -6,6 +6,7 @@ use std::pin::Pin;
>  use std::task::{Context, Poll};
>  
>  use futures::stream::{Stream, TryStream};
> +use futures::ready;
>  use tokio::io::{AsyncRead, AsyncWrite, ReadBuf};
>  use tokio::net::TcpListener;
>  use hyper::client::connect::Connection;
> @@ -108,10 +109,15 @@ impl hyper::server::accept::Accept for StaticIncoming {
>          self: Pin<&mut Self>,
>          cx: &mut Context,
>      ) -> Poll<Option<Result<Self::Conn, Self::Error>>> {
> -        match self.get_mut().0.poll_accept(cx) {
> -            Poll::Pending => Poll::Pending,
> -            Poll::Ready(Ok((conn, _addr))) => Poll::Ready(Some(Ok(conn))),
> -            Poll::Ready(Err(err)) => Poll::Ready(Some(Err(err))),
> +        let this = self.get_mut();
> +        loop {
> +            match ready!(this.0.poll_accept(cx)) {
> +                Ok((conn, _addr)) => return Poll::Ready(Some(Ok(conn))),
> +                Err(err) => {
> +                    eprintln!("error accepting connection: {}", err);
> +                    continue;
> +                }
> +            }
>          }
>      }
>  }
> -- 
> 2.20.1