[
https://issues.apache.org/jira/browse/GUACAMOLE-791?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16834804#comment-16834804
]
Michael Barkdoll commented on GUACAMOLE-791:
--------------------------------------------
Hi [[email protected]],
I've created a snip of the catalina.out log file during scenerio #1 from the
tomcat. Provided via docker command:
$ docker service logs -f -t --no-trunc --details vdi_guacamole2
[https://gist.github.com/michaelbarkdoll/6173fe6e2da0738a27a77266097cc025]
I also took a screenshot of my History page of connections. Interesting all my
connections seem to have the same Remote Host. I'm going to try and add one of
the machines to a VPN to see if that makes a difference with my reverse proxy
load balancing based upon ip_hash. I'll also see if there is some method of
increasing the verbosity of the logs.
> Connection Error from Balancing Connection Group
> ------------------------------------------------
>
> Key: GUACAMOLE-791
> URL: https://issues.apache.org/jira/browse/GUACAMOLE-791
> Project: Guacamole
> Issue Type: Bug
> Reporter: Michael Barkdoll
> Priority: Minor
> Attachments: Selection_999(2249).png, erro1.png, erro2.png
>
>
> I have a balancing connection group (cg1) with 2 VM's inside of it s1 and s2.
> I'm using three different machines (m1, m2, m3), one uses chrome the other
> two are firefox there is no profile sharing between the machines.
> => designates an action performed
> Scenerio 1:
> m1 => (user authenticates) test
> m2 => (user authenticates) test2
> m3 => (user authenticates) test3
> m1 => (user) test => (user initiates rdp connnection to) cg1 => (guacamole
> connection established to vm) s1
> m2 => (user) test2 => (user initiates rdp connnection to) cg1 => (guacamole
> connection established to vm) s2
> m3 => (user) test3 => (user initiates rdp connnection to) cg1 => Expectedly
> fails (all vms are in use)
> m2 => (user) test2 => (logoff rdp connection) s2
> m3 => (user) test3 => (user initiates rdp connnection to) cg1 => (guacamole
> connection established to vm) s2
> m3 => (user) test3 => (logoff rdp connection) s2
> m2 => (user) test2 => (user initiates rdp connnection to) cg1 => (guacamole
> connection established to vm) s2
> Everything appears to be working properly.
> Current active sessions:
> m1 [test][s1]
> m2 [test2][s2]
> But, now if I disconnect from s1 on machine m1, then for user test behavior
> becomes unexpected.
> m1 => (user) test1 => (logoff rdp connection) s1
> Current active sessions:
> m2 [test2][s2]
> Current available sessions:
> [s1]
> m3 => (user) test3 => (user initiates rdp connnection to) cg1 =>
> Connection Error: This connection is currently in use, and concurrent access
> to this connection is not allowed. Please try again later.
> m3 => (user) test3 => (user initiates rdp connnection to) cg1 => Reconnect
> link (nothing, wait 15sec, nothing)
> m3 => (user) test3 => (user initiates rdp connnection to) cg1 => Click Home,
> Wait 15sec
> m3 => (user) test3 => (user initiates rdp connnection to) cg1 =>
> Connection Error: This connection is currently in use, and concurrent access
> to this connection is not allowed. Please try again later.
> m3 => (user) test3 => (logoff guacamole .war frontend via AngularJS
> application)
> m3 => (user authenticates) test3
> m3 => (user) test3 => (user initiates rdp connnection to) cg1 => (guacamole
> connection established to vm) s1
> Current active sessions:
> m2 [test2][s2]
> m3 [test3][s1]
> Now things appear to be working for user test3, since the available s1 vm is
> assigned upon connecting ot the connection group.
> ------------
> Scenerio 2:
> machines: m1, m2, m3
> vms: s1, s2 (in balancing connection group, cg1)
> m1 => auth test
> m2 => auth test2
> m2 => auth test3
> m1 => user test => cg1 => s1
> m2 => user test2 => cg1 => s2
> m2 => user test2 => logoff vm s2
> m3 => user test3 => cg1 => s2
> m3 => user test3 => logoff vm s2
> m2 => user test2 => cg1 => s2
> Current active sessions:
> m1 [test][s1]
> m2 [test2][s2]
> # Logoff s1 (s2 is still active on m2)
> m1 => user test1 => logoff vm s1
> Current active sessions:
> m2 [test2][s2]
> Current available sessions:
> [s1]
> m3 => user test3 => cg1 =>
> Connection Error: This connection is currently in use, and concurrent access
> to this connection is not allowed. Please try again later.
> Ok, that was expected from my previous experience, but now logoff the last
> active session on m2:
> m2 => user test2 => logoff s2
> Current active sessions:
> none
> Current available sessions:
> [s1][s2]
> Now attempt to reconnect with m3, you'd think you'd get s1 with least number
> of connection balancing, but no its s2.
> Wait 15 second (for api sync?)
> m3 => user test3 => cg1 => s2
> Current active sessions:
> m3 [test3][s2]
> Current available sessions:
> [s1]
> Now attempt to connect m2,
> m2 => user test2 => cg1 =>
> Connection Error: This connection is currently in use, and concurrent access
> to this connection is not allowed. Please try again later.
> Attempt to connect m1,
> m1 => user test => cg1 => s1 (works)
> Current active sessions:
> m1 [test][s1]
> m3 [test3][s2]
> I have stickiness enabled on the nginx could that be related?
> https://github.com/michaelbarkdoll/guacamole-reverse-proxy-nginx/blob/master/nginx.conf
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
