tramp-devel
[Top][All Lists]
Advanced

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

Re: ssh through a bastion


From: Michael Albinus
Subject: Re: ssh through a bastion
Date: Sat, 01 Feb 2020 13:18:19 +0100
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/28.0.50 (gnu/linux)

Garjola Dindi <address@hidden> writes:

> Hi,

Hi Garjola,

[Pls check your mail settings. Your message arrived as base64 encoded
chunk, which I have decoded manually. For convenience of other people on
this ML, I keep a fullquote]

> I am using tramp to ssh to a cluster which uses a bastion (I think this
> is the name for this thing!). Usually, to connect to the cluster, I do
>
> ssh user@cluster
>
> I asks for a password, and presents a curses menu to select the type of
> node (after a timeout it will select a default option). After that, it
> asks again for the password and connects to a node of the cluster. In
> the case of scp, there is no curses menu, but I still have to enter the
> password twice.
>
> I can use tramp without problem, since the timeout will select the
> default node, which is OK for me.
>
> However, I have to enter the password twice, which is inconvenient.
> Also, from time to time, it will ask for the password again (and twice
> every time).

Hmm. It would be good to know the dialogue between Tramp and that
machine.

> I thought that using the .authinfo file I would not need to enter the
> password, but it still asks for it (but only once).

For the first password, Tramp understands that it belongs to the
machine, and it (re-)uses the saved value from .authinfo.

> I think this is different from the multi-hop that I have read in the
> manual, since the second connection does not ask for a login (it is
> the same as the one of the bastion) and I don't get to choose the
> machine (cluster node) to which I am connecting.

Exactly.

> Is there a way to avoid to manually entering the password in this kind
> of situation?

Pls set tramp-verbose to 6, and rerun the test. There will be a Tramp
debug buffer, which shall tell us more. This debug buffer won't contain
any password in clear text, but you might double-check.

> I hope my description above is not too vague and someone can help me.

At least I have an idea what's up :-)

> Thank you.
>
> G.

Best regards, Michael.



reply via email to

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