Control: tag -1 patch
Hi,
Putting the attached file into your "extensions/post-suspend.d" folder on the server... may solve #302 issues...
On Ubuntu that would be an executable named "302_bug-fix" with the following full path: /usr/lib/x2go/extensions/post-suspend.d/302_bug-fix (On other distros just simply locate your "post-suspend.d" folder and put it there.. PS! Make sure its executable! (chmod 755).
Its may be good to clean up all old sessions before you start a brand new and fresh session to test this fix.
We were able to reliably reproduce fault/non-fault conditions and after applying this fix, we're seeing 100% success rates for session resume with pyhoca....
The source of the issue seem to be that the NX port reappears on the server within seconds of session suspension.... in a state where where its not ready for new incoming connections... (quite frankly it should not be available at all at this time, should it?) So either we're reconnecting to this "faulty" port or the fresh port fails to open due to the "old port" being in the way.... Either way NX fails to resume...
The problem seem to be consistently occurring when rapidly suspending/resuming with pyhoca and if x2goclient experience abnormal/unclean disconnects that results in suspension.
Either way this little perl script will clean up those stray NX ports by simply locating and connecting to them... (after which, the port closes, as it should.)
If this indeed solves the #302 bug, this should probably move from extensions to being part of the suspend scripts.... but for cleanliness and admin-friendliness its easier to test it as an extension....
Let us know if this works for you!
(btw: If the supportive subs are adopted to Utils.pm the script it self would become quite compact)
Best Regards, -GZNGET