tramp-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ssh ControlMaster=yes, known_hosts, etc.


From: Michael Albinus
Subject: Re: ssh ControlMaster=yes, known_hosts, etc.
Date: Sat, 10 Apr 2010 16:46:50 +0200
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/24.0.50 (gnu/linux)

Gary <address@hidden> writes:

>> > I'm asuming it does (when using ssh) but I've occasionally seen odd
>> > behaviour that makes me wonder.
>>
>> What, exactly ?
>
> Exactly? I'm not sure I remember exactly *laughs* I do remember thinking
> a few times "It should have that password already cached, why am I being
> asked for it again?" (I have password-cache-expiry set to nil.) That may
> be what I was thinking of.

Sometimes, when Tramp looses a connection, it clears all cached
connection information, including passwords. Tramp cannot know the
reason for a connection loss, so it might be better to start with fresh
values.

>> There's a per-host tramp buffer which 'caches' the connection. If you
>> kill it, the ssh process goes too.
>
> Yeah, I'd seen that, but IIRC killing that buffer didn't help either. I
> will check next week, but I can't right now. Is there a reason the
> killing of the ssh process depends on that buffer and no the buffer
> relating to the file? I can understand it when sharing the ssh
> connection, you don't want the connection to die just because you close
> one file of a dozen open on one particular host, but I don't, and indeed
> can't.

An Emacs process is always bound to one buffer, and so does
Tramp. Killing that buffer shall also kill the corresponding process
(but you are on MS Windows; dunno whether this is also true there).

OTOH, killing that buffer is often not sufficient. Tramp keeps also
cached information, which must be flushed as well. Calling
tramp-cleanup-connection or tramp-cleanup-all-connections is safer in
this respect.

> Is the "per-host tramp buffer" (PHTB) really per host, even if I am
> using one ssh process per connection?

One process per connection. A connection is identified by method+user+host.

> I have to say, I find the presence of the PHTBs kind of intrusive,
> since I often have many buffers open, and they just add to the
> noise. I could use ido-ignore-buffers to hide them, but if I need to
> find 'em to kill the processes that doesn't seem like a terrific idea.

This I do not understand. Could you, please, explain in detail what
happens to you?

Best regards, Michael.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]