FreeTDS Development Group

[Agnieszka Figiel <agnieszka AT llp.pl>]

2011/6/28 Frediano Ziglio <freddy77 AT gmail.com>:
> 2011/6/28 Agnieszka Figiel <agnieszka AT llp.pl>:
>> 2011/6/28 Frediano Ziglio <freddy77 AT gmail.com>:
>>> 2011/6/28 Agnieszka Figiel <agnieszka AT llp.pl>:
>>>> The error encountered during tsql session error is as follows:
>>>>
>>>> Error 20004 (severity 9):
>>>>       Read from the server failed
>>>>       OS error 104, "Connection reset by peer"
>>>>
>>>
>>> I checked error 104 and is ECONNRESET (as string correctly reported).
>>> From log we send data and server refure it. Perhaps it does not like
>>> something in gnutls encoding :(
>>
>> You were right, apparently it works --with-openssl! That's really
>> great news, thank you!
>>
>> However, I'm still kind of confused about why it worked, maybe you
>> have a clue :) When we first started with freetds we used 0.82 from
>> Ubuntu package -- am I right that it comes compiled --with-openssl by
>> default? I'm quite certain it did not work for me at that time (same
>> db, same machine), I even posted about this to this list; problem was
>> resolved when I compiled --with-gnutls. So I guess 'something' must
>> have changed between the freetds versions that affected this -- just
>> curious if you have a guess what that could be?
>>
>> Thanks a lot!
>>
>
> Hi,
>  could you try this patch with GnuTLS ??
>
> diff --git a/src/tds/net.c b/src/tds/net.c
> index 894ab4e..b060ff8 100644
> --- a/src/tds/net.c
> +++ b/src/tds/net.c
> @@ -1331,6 +1331,7 @@ tds_ssl_init(TDSSOCKET *tds)
>                gnutls_compression_set_priority(session, comp_priority);
>                gnutls_kx_set_priority(session, kx_priority);
>                gnutls_mac_set_priority(session, mac_priority);
> +               gnutls_record_disable_padding(session);
>
>                /* put the anonymous credentials to the current session */
>                tls_msg = "setting credential";
>
> I'm unable to reproduce your problem. I tried to send very long
> queries but my GnuTLS version keep sending only 0x1010 bytes per
> packet (patch applied or not).

Hello,

you're right again, this patch works for my case! Thanks!

-- 
Agnieszka