Re: cannot quit: stale NFS handle

[Bruno H. <br...@cl...>]

Sam wrote:
> > Did you try unmounting and remounting the mount? Does it help? I.e. is
> > ESTALE a temporary condition, or does it persist across remount?
>
> This is solaris.  I am not root.

Well, in the absence of factual knowledge, I assume that ESTALE is a
temporary condition. Therefore the user has the choice between
  a) remounting and continuing successfully,
  b) losing data and throwing away the stream.

Since one of the options implies "losing data", we shouldn't make the
choice automatic.

> maybe close_all_files() should set some error handler.
> (i.e., wrap everything in an equivalent of IGNORE-ERRORS but print the
> name of the file it could not close).

That would be fine... if ignoring the errors would not lose data.

But the way things are, it's probably better to install a restart, not
a handler, with 2 options: a) retry, b) force closing the stream even if
it loses data. This handler should be established inside the
builtin_stream_close function.

Bruno