[pve-devel] [PATCH http-server 1/1] webproxy: handle unflushed write buffer
Fabian Grünbichler
f.gruenbichler at proxmox.com
Mon Nov 8 16:45:19 CET 2021
On November 8, 2021 3:15 pm, Fabian Ebner wrote:
> Am 05.11.21 um 14:03 schrieb Fabian Grünbichler:
>> for proxied requests, we usually tear down the proxy connection
>> immediately when closing the source connection. this is not the correct
>> course of action for bulk one-way data streams that are proxied, where
>> the source connection might be closed, but the proxy connection might
>> still have data in the write buffer that needs to be written out.
>>
>> push_shutdown already handles this case (closing the socket/FH after it
>> has been fully drained).
>>
>> one example for such a proxied data stream is the 'migrate' data for a
>> remote migration, which gets proxied over a websocket connection.
>> terminating the proxied connection early makes the target VM crash for
>> obvious reasons.
>>
>> Signed-off-by: Fabian Grünbichler <f.gruenbichler at proxmox.com>
>> ---
>> src/PVE/APIServer/AnyEvent.pm | 10 ++++++----
>> 1 file changed, 6 insertions(+), 4 deletions(-)
>>
>> diff --git a/src/PVE/APIServer/AnyEvent.pm b/src/PVE/APIServer/AnyEvent.pm
>> index 86d0e2e..ecf771f 100644
>> --- a/src/PVE/APIServer/AnyEvent.pm
>> +++ b/src/PVE/APIServer/AnyEvent.pm
>> @@ -144,7 +144,8 @@ sub client_do_disconnect {
>> };
>>
>> if (my $proxyhdl = delete $reqstate->{proxyhdl}) {
>> - &$shutdown_hdl($proxyhdl);
>> + &$shutdown_hdl($proxyhdl)
>> + if !$proxyhdl->{block_disconnect};
>
> Style nit: fits in one line ;)
>
> I'm not familiar with the code, so I'll just ask: can this be reached
> without going through the code below first, i.e. can it happen that
> block_disconnect is not set, but length $proxyhdl->{wbuf} > 0? Or is it
> not important in other cases (if there are any)?
in theory, yes. in practice, not likely with anything that matters ;)
we have
- spiceproxy, if the client closes the connection we likely
don't care about the last few bytes of input being dropped
- proxied API requests with response_stream - never gets written to
- WS proxy (fixed in this patch ;) technically also used like
spiceproxy)
but I have to admit I did think about doing this always and relying on
timeouts to clear the proxyhdl eventually, but then decided to play it
safe and just change the one I am sure requires it..
>
>> }
>>
>> my $hdl = delete $reqstate->{hdl};
>> @@ -627,9 +628,10 @@ sub websocket_proxy {
>> } elsif ($opcode == 8) {
>> my $statuscode = unpack ("n", $payload);
>> $self->dprint("websocket received close. status code: '$statuscode'");
>> - if ($reqstate->{proxyhdl}) {
>> - $reqstate->{proxyhdl}->push_shutdown();
>> - }
>> + if (my $proxyhdl = $reqstate->{proxyhdl}) {
>> + $proxyhdl->{block_disconnect} = 1 if length $proxyhdl->{wbuf} > 0;
>> + $proxyhdl->push_shutdown();
>> + }
>> $hdl->push_shutdown();
>> } elsif ($opcode == 9) {
>> # ping received, schedule pong
>>
>
More information about the pve-devel
mailing list