Hello,
If you look at my log in the two cases when the WoL is active i get
guacd[2674]: DEBUG: Sending Wake-on-LAN packet, and pausing for 15
seconds.
guacd[2674]: DEBUG: Client is using protocol version "VERSION_1_5_0"
guacd[1]: INFO: Connection "$6e922686-03ca-43ac-82f2-d1c3079d8074"
removed.
guacd[1]: DEBUG: Unable to request termination of client process: No
such process
guacd[1]: DEBUG: All child processes for connection
"$6e922686-03ca-43ac-82f2-d1c3079d8074" have been terminated.
Aren't the last two entries a signal of something going wrong inside guacd?
In fact that explains why even if the machine is up and running there is no
way to establish a connection as long as the WoL tick is present
Regards,
Il giorno dom 24 dic 2023 alle ore 21:28 Nick Couchman <vn...@apache.org>
ha scritto:
> On Sun, Dec 24, 2023 at 5:06 AM Massimiliano Gilli <max90g...@gmail.com>
> wrote:
>
>> Hello,
>> im playing around with guacamole on docker coupled with virtualbox (VRDE)
>> The connection works fine and I can browse my VMs as long as they are
>> powered on.
>> To overcome the fact that they must be powered on, I started to play
>> around with the WoL feature.
>> I was successfully able to receive WoL packets from the guacamole
>> frontend container to my host machine and via tcpdump I'm able to listen
>> for these packets so that I can trigger a "VBoxManage startvm" command the
>> poweron a vm.
>> The issue is that even if the VM comes up and even with 5 seconds of Host
>> boot wait time the connection is not established.
>> I think i'm quite close to a decente result because if immediately after
>> the first connection request (which starts up the vm) i disable the WoL
>> feature from guacamole frontend and reconnect i can see the boot logo and
>> work on it normally.
>> Is the WoL feature expecting a reply from the machine that is woken up
>> before trying to connect?
>>
>>
> No, the WoL feature does not wait for a reply from the target machine - it
> just sends the WoL packet and then attempts the connection. This generally
> results in a connection "failure" the first time around, and then success,
> depending on how long it takes the VM to boot. There is a feature request
> out there somewhere in Jira to make the WoL feature a little bit more
> "intelligent" in this regard, like sending the WoL packet and then pinging
> or attempting a TCP connect to the machine that has just been woken up
> until it responds so that you don't get the failure, timeout, and then have
> to retry.
>
> It is worth nothing that you do not need to disable the WoL feature before
> re-trying the connection - it should just work as soon as the system is
> available.
>
> -Nick
>
>>
