Ubuntu 16.04.5 – Console display access unavailable

[khawkins.sfdc]

I’m running into an issue similar to the one posted in https://forums.nomachine.com/topic/errors-with-6-0-78-on-ubuntu-16-04.  Namely, if I establish a remote session connection to my Ubuntu server (from an OS X client, FWIW) while I’m not logged into the physical console, I notice two things:

1. An option to establish a connection to the physical console session is not available.
2. /usr/NX/var/db/failed and /usr/NX/var/log/node are continuously spawning new entries without the old ones being cleaned up, eventually running my root partition out of inodes.

The following log lines keep showing up every 1-2 seconds in /usr/NX/var/log/nxserver.log (which itself has grown to 120MB, as of this writing):

2018-10-12 15:50:28 419.668 21687 NXSERVER WARNING! Process ‘/usr/NX/bin/nxexec –node –user lightdm –priority realtime –mode 0 –pid 25’ with pid ‘64031/64031’ finished with exit code 11 after 1,684 seconds.

2018-10-12 15:50:28 419.969 21687 NXSERVER WARNING! 9C578F29A5DC4F700A24799672CFEE03 nxnode died with exit code 11.

2018-10-12 15:50:28 558.566 21687 NXSERVER WARNING! Session 9C578F29A5DC4F700A24799672CFEE03 crashed on display 0.

The following errors keep recurring in nxerror.log, though I don’t know that they’re related:

72494 72494 15:59:31 259.680 Socket: WARNING! Can’t bind the IPv6 socket.

72494 72494 15:59:31 259.740 Socket: WARNING! Error is 99 ‘Cannot assign requested address’.

Establishing virtual sessions still works fine.

Both NX server and client are running NoMachine 6.2.4.  The server is running NoMachine Workstation Subscription Pack, while the client is the free version. It seems weird that, if a physical console session has been established previously (by logging in at the console), everything’s fine.  But NoMachine otherwise seems unable to initiate that process itself, if it hasn’t already been established.  Any clues / insights would be most appreciated.

Server: Ubuntu 16.04.5 LTS

Client: macOS High Sierra (10.13.6)

NoMachine Server: 6.2.4 Workstation

NoMachine Client: 6.2.4 free client

[graywolf]

Could you create an archive of folder /usr/NX/var/log ?

Could you check if any crash file is created in /var/crash?

Please, send everything to the email address forum[at]nomachine[dot]com.

[khawkins.sfdc]

I’ve sent an email with the contents of /usr/NX/var/log.  There was nothing to report from /var/crash.  Let me know if I can provide further details.

[Britgirl]

We received the logs and will provide some feedback soon.

[graywolf]

Could you check if any crash is reported by the command

sudo apport-cli

May I look at the lightdm process tree when nobody is logged in the physical console? This command should be able to dump it:

pstree $(ps --ppid 1 | grep lightdm | awk '{print $1}')

[khawkins.sfdc]

Thanks. I sent those along via email.  There was a crash log, which looks like it originated in /usr/NX/bin/nxnode.bin.  Hopefully the information provided there will be useful.

[graywolf]

Likely we found the source of the problem. We’ll send you a library that should fix it.

[graywolf]

Update.

The problem looks fixed in the latest version (6.3.6). Would you upgrade rather than patching v. 6.2.4 ?

[khawkins.sfdc]

I installed 6.3.6 Workstation and the 6.3.6 client on my OS X machine, and I’m happy to report that, at least as of +3 hours of work with a remote session, I’m no longer seeing the crashing behavior!  I still can’t connect to the console session when I launch my NX client.  But that was always less of a concern to me than the server software crashing and ultimately running my partition out of inodes.  Thanks so much for your help!  I’ll post any updates I have, if things regress, etc.

[graywolf]

You need to add the “trusted” qualification to your user profile in order to get access to the login screen. Try this:

/etc/NX/nxserver --useredit <username> --trusted physical

[Author]

Posts