• SVDM - strange issue "already connected"

    From deon@ALTERANT to Digital Man on Fri Aug 4 17:23:21 2023
    Hey DM,

    Not sure if you recall, but earlier this year I was having some problems with SVDM where after a while it would no longer connect.

    Not sure if this is related, but I discovered a repeatable issue today.

    In my svdm.ini file, I have an alias of

    010003001=c-8-2.dege.lan:11179

    That host resolves to both an IPv4 and IPv6 address. I think windows is prefering the IPv6 address when the resolver present both (although I'm not 100% sure if that is correct), but when I replace the hostname with an IPv4 address - the problem below does *not* occur. (So its either an IPv6 thing, or a resolver thing.)

    Anyway, if when my (dos) mailer is calling to my uplink, and it gets a connection refused, the mailer (via SVDM) reports "NO CARRIER" (which I think is expected).

    However, the next attempt results in SVDM reporting "ERROR" - regardless of whether the uplink is back online or not. If I try and call in to the mailer, I get SVDM responding "Sorry, not available right now".

    Using dbgview, I can see each attempt.

    The first failure:
    00000049 5:08:15 PM [3472] SBBSVDM: RX command: ' ATD010003001 ' at 3035779
    00000050 5:08:15 PM [3472] SBBSVDM: dial(010003001)
    00000051 5:08:15 PM [3472] SBBSVDM: Connecting to port 11179 at host 'c-8-2.dege.lan' via Raw
    00000052 5:08:15 PM [3472] SBBSVDM: C:\GitLab-Runner\builds\4S9KMgUy\2\main\sbbs\src\vdmodem\vdmodem.c 640 calling getaddrinfo
    00000053 5:08:15 PM [3472] SBBSVDM: C:\GitLab-Runner\builds\4S9KMgUy\2\main\sbbs\src\vdmodem\vdmodem.c 662 calling connect
    00000054 5:08:15 PM [3472] SBBSVDM: Setting socket options
    00000055 5:08:15 PM [3472] SBBSVDM: Socket send buffer length: 8192 bytes
    00000056 5:08:15 PM [3472] SBBSVDM: Socket receive buffer length: 8192 bytes
    00000057 5:08:15 PM [3472] SBBSVDM: Socket TCP_NODELAY: 0
    00000058 5:08:15 PM [3472] SBBSVDM: C:\GitLab-Runner\builds\4S9KMgUy\2\main\sbbs\src\vdmodem\vdmodem.c 710 socket_recvdone
    00000059 5:08:15 PM [3472] SBBSVDM: Modem response: NO CARRIER

    The subsequent failure:

    00000070 5:08:23 PM [3472] SBBSVDM: RX command: ' ATD010003001 ' at 3043682
    00000071 5:08:23 PM [3472] SBBSVDM: Can't dial: Already connected 00000072 5:08:23 PM [3472] SBBSVDM: Modem response: ERROR

    IE: It is reporting "already connected" (when it isnt).

    The only way I can clear this is unload SVDM and reload it and then I'm good to go until next time the uplink (only 1 uplink tested) is offline.

    Thought you might like to know.

    (This is SVDM from 2a6d9983)


    ...δεσ∩

    ---
    ■ Synchronet ■ AnsiTEX bringing back videotex but with ANSI
  • From Digital Man to deon on Fri Aug 4 12:17:08 2023
    Re: SVDM - strange issue "already connected"
    By: deon to Digital Man on Fri Aug 04 2023 05:23 pm

    Hey DM,

    Not sure if you recall, but earlier this year I was having some problems with SVDM where after a while it would no longer connect.

    Not sure if this is related, but I discovered a repeatable issue today.

    In my svdm.ini file, I have an alias of

    010003001=c-8-2.dege.lan:11179

    That host resolves to both an IPv4 and IPv6 address. I think windows is prefering the IPv6 address when the resolver present both (although I'm not 100% sure if that is correct), but when I replace the hostname with an IPv4 address - the problem below does *not* occur. (So its either an IPv6 thing, or a resolver thing.)

    Anyway, if when my (dos) mailer is calling to my uplink, and it gets a connection refused, the mailer (via SVDM) reports "NO CARRIER" (which I think is expected).

    However, the next attempt results in SVDM reporting "ERROR" - regardless of whether the uplink is back online or not. If I try and call in to the mailer, I get SVDM responding "Sorry, not available right now".

    Using dbgview, I can see each attempt.

    The first failure:
    00000049 5:08:15 PM [3472] SBBSVDM: RX command: ' ATD010003001 ' at 3035779
    00000050 5:08:15 PM [3472] SBBSVDM: dial(010003001)
    00000051 5:08:15 PM [3472] SBBSVDM: Connecting to port 11179 at host 'c-8-2.dege.lan' via Raw
    00000052 5:08:15 PM [3472] SBBSVDM: C:\GitLab-Runner\builds\4S9KMgUy\2\main\sbbs\src\vdmodem\vdmodem.c 640 calling getaddrinfo
    00000053 5:08:15 PM [3472] SBBSVDM: C:\GitLab-Runner\builds\4S9KMgUy\2\main\sbbs\src\vdmodem\vdmodem.c 662 calling connect
    00000054 5:08:15 PM [3472] SBBSVDM: Setting socket options 00000055 5:08:15 PM [3472] SBBSVDM: Socket send buffer length: 8192 bytes
    00000056 5:08:15 PM [3472] SBBSVDM: Socket receive buffer length: 8192 bytes
    00000057 5:08:15 PM [3472] SBBSVDM: Socket TCP_NODELAY: 0 00000058 5:08:15 PM [3472] SBBSVDM: C:\GitLab-Runner\builds\4S9KMgUy\2\main\sbbs\src\vdmodem\vdmodem.c 710 socket_recvdone
    00000059 5:08:15 PM [3472] SBBSVDM: Modem response: NO CARRIER

    The subsequent failure:

    00000070 5:08:23 PM [3472] SBBSVDM: RX command: ' ATD010003001 ' at 3043682
    00000071 5:08:23 PM [3472] SBBSVDM: Can't dial: Already connected 00000072 5:08:23 PM [3472] SBBSVDM: Modem response: ERROR

    IE: It is reporting "already connected" (when it isnt).

    Thank you! That is very helpful.

    The only way I can clear this is unload SVDM and reload it and then I'm good to go until next time the uplink (only 1 uplink tested) is offline.

    Looking at the code, this is a bug, but I also think an "ATH0" would have cleared this condition.

    Thought you might like to know.

    (This is SVDM from 2a6d9983)

    Thanks. I'll commit a fix in a second here, but nothing specifically for IPv6. Let me know if there remains some issue with the IPv6 support and I'll look into that.
    --
    digital man (rob)

    This Is Spinal Tap quote #11:
    Nigel Tufnel: No. no. That's it, you've seen enough of that one.
    Norco, CA WX: 81.2°F, 49.0% humidity, 5 mph E wind, 0.00 inches rain/24hrs
  • From deon@ALTERANT to Digital Man on Sat Aug 5 16:33:13 2023
    Re: SVDM - strange issue "already connected"
    By: Digital Man to deon on Fri Aug 04 2023 12:17 pm

    Hey DM,

    Thanks. I'll commit a fix in a second here, but nothing specifically for IPv6. Let me know if there remains some issue with the IPv6 support and I'll look into that.

    Thanks, that seems to have fixed it (using a full hostname at least). I'll leave it running and see how it goes. I havent tested separating out IPv4/IPv6 hostnames (not really sure why that would be an issue) - but if I have problems, I'll try and get a repeatable problem.


    ...δεσ∩

    ---
    ■ Synchronet ■ AnsiTEX bringing back videotex but with ANSI