On Fri, Jun 7, 2019 at 5:12 AM Vieri <[email protected]> wrote:

> Hi,
>
> In a previous message I mentioned a problem I was seeing with several
> Firefox versions. Whenever I try to connect to either a telnet service or
> RDP server via Guacamole, I sometimes get an odd behavior as if the user
> (me) were sending out ENTER keystrokes (I sent a screenshot). This is
> usually fixed by logging out and back in again. However, it can randomly
> occur again until I log out once again.
>
> This is what I get in my logs when it happens.
>
> STDOUT:
>
> 13:58:06.442 [https-openssl-apr-8443-exec-7] DEBUG
> o.a.g.a.ldap.LDAPConnectionService - Connection to LDAP server without
> encryption.
> 13:58:06.494 [https-openssl-apr-8443-exec-7] DEBUG
> o.a.g.auth.ldap.ObjectQueryService - Searching
> "cn=Users,dc=mydomain,dc=org" for objects matching
> "(&(!(objectClass=guacConfigGroup))(member=cn=myuser,cn=Users,dc=mydomain,dc=org))".
> 13:58:06.812 [https-openssl-apr-8443-exec-7] INFO
> o.a.g.r.auth.AuthenticationService - User "myuser" successfully
> authenticated from 10.215.144.48.
> 13:58:06.813 [https-openssl-apr-8443-exec-7] DEBUG
> o.a.g.a.ldap.LDAPConnectionService - Connection to LDAP server without
> encryption.
> 13:58:06.833 [https-openssl-apr-8443-exec-7] DEBUG
> o.a.g.auth.ldap.ObjectQueryService - Searching
> "cn=Users,dc=mydomain,dc=org" for objects matching
> "(&(|(objectClass=guac*)(cn=*IT*)(cn=*VPN*)(cn=*myuser*))(cn=*))".
> 13:58:07.116 [https-openssl-apr-8443-exec-7] DEBUG
> o.a.g.auth.ldap.ObjectQueryService - Searching
> "cn=Users,dc=mydomain,dc=org" for objects matching
> "(&(!(objectClass=guacConfigGroup))(cn=*))".
> 13:58:07.963 [https-openssl-apr-8443-exec-7] WARN
> o.a.g.auth.ldap.ObjectQueryService - Failed to process an LDAP search
> result. Error was: Sizelimit Exceeded
> ...
>

You'll probably want to look into your LDAP query limits if you're
expecting listing of LDAP users within guac's admin interface to succeed,
but this will have no bearing on handling of keyboard events. The two parts
of the stack which could contain a problem resulting in a mishandled event
are:

1) The browser-side keyboard event handling within JavaScript
2) If keyboard events within the browser are handled correctly, the
handling of received events within protocol-specific event handling
(libguac-client-rdp, etc.)

RDP keyboard event translation is complex, but if you're seeing this same
behavior regardless of underlying protocol, and specifically within
Firefox, that suggests something may be behaving differently on the
JavaScript side.

We have a test page for just the keyboard event handling which can be used
to verify whether things are not behaving as expected in the client. Try
testing against that with the affected browser and see whether you see the
same issue:

http://guacamole.apache.org/pub/tests/guac/keyboard-test.html

...
> Jun  7 13:58:11 inf-fw3 guacd[18791]: Internal RDP client disconnected
> Jun  7 13:58:11 inf-fw3 guacd[14495]: Connection
> "$ae4363ce-1671-40f0-8734-ce45d6af39ad" removed.
> Jun  7 13:58:11 inf-fw3 kernel: guacd[18797]: segfault at 7fd7ea22001c ip
> 00007fd802aafb8b sp 00007fd801ccece0 error 4 in
> libguac-client-rdp.so.0.0.0[7fd802a98000+2c000]
>
>
> Note the segfault.
>
> What can I try?
> Is it worth a JIRA report?


Segfaults within the RDP support tend to be due to older builds of the
guac*.so plugins for FreeRDP hanging around from older builds of
guacamole-server, so I wouldn't recommend opening a new JIRA issue at this
point. First verify that there is nothing hanging around from older builds
(or from distribution packages, or third-party packages). If there is any
chance that the FreeRDP libraries may have been upgraded since you last
built things, I'd recommend rebuilding as FreeRDP releases are not
generally API- nor ABI-compatible, and this could also cause a segfault
without there being an issue within Guacamole.

If you can confirm the above, and the segfault is reproducible, a backtrace
from gdb would be the next step to determine what is happening. I don't
believe this is related to the keyboard behavior you're seeing.

- Mike

Reply via email to