Re: Nautilus hangs on long operations

[Tomas Bzatek <tbzatek redhat com>]

On Wed, 2010-12-01 at 16:11 +0000, Rui Tiago Cação Matos wrote:
> 2010/12/1 Pierre Wieser <pwieser trychlos org>:
> > Mostly all my work is made on a NFS server (same configuration)
> > via a gigabit LAN.
> 
> Word on the street is that NFS is crap. I put my money on that being
> your problem right there, no matter how fast your LAN is. That being
> said, I never used NFS and I don't know how specially, if at all, such
> mounts are treated by GIO/gvfs/nautilus.

There are no special quirks for NFS except of file monitoring, which is
done by polling. From my experience large traffic on NFS causes lags,
sometimes 4-5 seconds. So even a simple statfs() call used for updating
free space on a volume might cause the process to stuck. Not sure if all
these trivial operations are async.

NFS mounts can be heavily tuned, see the manpages. Also, not sure how
kernel handles I/O workload queuing, if ever.

I wonder if you can reproduce this on local filesystems. Large I/O load
on local disks can cause slowdowns of whole desktop, not just Nautilus
specifically.

-- 
Tomas Bzatek <tbzatek redhat com>