[Thinlinc-technical] very unstable mechanism to share client-side media via NFS-mounts

[Peter Astrand]

Hi and thanks for your feedback. We are aware of that these types of 
problems happens from time to time. The problem is actually not the use of 
the NFSv3 protocol, but the fact that we are using the standard Linux 
kernel NFS client, which is not well suited for this use case. This is 
also an area with constant development, so the behaviour (amount of 
problems) may depend on which kernel you use (ie affected by choice of 
distribution).

Have you tried "tl-umount-localdrives -vas"? Does it help? It may take 
some time, but it should eventually succeed.

Which Linux kernel & distribution are you running?

For the future, we are considering change the server side to use a FUSE 
based NFS client, instead of the kernel client:

https://www.cendio.com/bugzilla/show_bug.cgi?id=4900

It could also be interesting to hear what kind of client platforms you are 
using.

Best regards, 
Peter

On Tue, 29 Nov 2016, Rob De Langhe wrote:

> 
> hi,
> 
> I must say that the mechanism (up till TL-version 4.7) by which client-side media (hard disk, USB-stick, ...) are 'exported' via
> NFS to the TL-agents, is quite unstable.
> 
> There is no bullet-proof procedure by which logout scripts can avoid that these NFS-mounts become 'stale'. Result is that the
> TL-agent can only recover from such 'stale' NFS mounts via a reboot. Or worse: if the TL-agent is a *NIX container then its 'guest'
> server (who runs the kernel) needs a reboot as well...
> 
> -> is there no better mechanism to share the client-side media via the VNC client-to-agent tunnels? Another protocol than NFS is
> required to avoid the above problems.
> 
> brgds
> Rob
> 
> 
>


---
Peter Astrand		ThinLinc Chief Developer
Cendio AB		https://cendio.com
Teknikringen 8		https://twitter.com/ThinLinc
583 30 Linkoping	https://facebook.com/ThinLinc
Phone: +46-13-214600	https://google.com/+CendioThinLinc