On Wed, Apr 7, 2021 at 11:11 PM Thomas Besser <[email protected]> wrote:
> Am 08.04.21 um 00:02 schrieb Mike Jumper: > > On Wed, Apr 7, 2021 at 7:57 AM Thomas Besser <[email protected] > > <mailto:[email protected]>> wrote: > > Obviously "wol-wait-time" does not really "wait". How can I solve > this > > issue? > > > > It should. There is a 15 second network timeout, yes, but Guacamole > > should be sending messages back and forth while it waits for WoL, > > keeping the connection alive despite the remote desktop not yet being > > available. If you set "wol-wait-time" to "40", then the connection > > process should wait at least 40 seconds before giving up. > > > > What protocol is being used to connect to the remote desktop in this > case? > > RDP > > Here a log sequence from guacd: > > 2021-04-08T06:02:33.205499219Z guacd[2706]: DEBUG: Client is using > protocol version "VERSION_1_3_0" > 2021-04-08T06:02:33.205533781Z guacd[2706]: DEBUG: Sending > Wake-on-LAN > packet, and pausing for 40 seconds. > 2021-04-08T06:02:48.221240007Z guacd[2706]: ERROR: User is not > responding. > 2021-04-08T06:02:48.222878202Z guacd[2706]: INFO: User > "@69dc9288-c346-4701-ab35-1ebca1ae2ea6" disconnected (0 users remain) > 2021-04-08T06:02:48.223000371Z guacd[2706]: INFO: Last user of > connection "$a55c90a9-5a75-4b5e-98d1-55973550523b" disconnected > 2021-04-08T06:02:48.223235827Z guacd[2706]: DEBUG: Requesting > termination of client... > 2021-04-08T06:02:53.224247047Z guacd[2706]: WARNING: Client did not > terminate in a timely manner. Forcibly terminating client and any child > processes. > 2021-04-08T06:02:53.232161722Z guacd[7]: INFO: Connection > "$a55c90a9-5a75-4b5e-98d1-55973550523b" removed. > > BTW it's a docker installation behind a proxy. > > As mentioned in another mail: I'm using 'balancing connection group'. > Only if "enable session affinity" is activated it does work generally > (with connection errors). If this option is disabled, it is looping and > never connects to the remote host, even it is running meanwhile. > If you connect directly to the connection (without a balancing group at all), does the connection wait for the expected 40 seconds? There is internal handling of connection failures within balancing groups that is intended to automatically and transparently fail over to another connection within the group if the current one explicitly fails. It may be that this functionality is inadvertently eating the keep-alive messages sent during the WoL wait, causing the connection to time out. Michael Jumper CEO, Lead Developer Glyptodon Inc <https://glyp.to/>.
