[Sciserver-users] SciServer new user questions
Tom Dwelly
dwelly at mpe.mpg.de
Tue Oct 20 18:08:38 CEST 2020
Hi Max,
yes I have been using the web-based terminal interface already. Unfortunately the subtle differences between that
interface and the 'real' terminal I'm used to are a bit of a barrier. But it's only day one of trying this out, so maybe
my muscle memory will give in eventually. I'm investigating firefox extensions that will enable 'normal' linux-like copy
and paste operations in the web-terminal.
However, I was mainly interested in the additional functionality that would be enabled by being able to pipe stuff in
and out of a sciserver container via an ssh tunnel. But it sounds like that's not possible. Oh well, there are other routes.
Cheers,
Tom
On 20/10/2020 15:53, Maximilian Fabricius wrote:
> Hi Tom,
>
> (good to hear from you virtually again)
>
> ... and you do know that there is a terminal interface in SciServer, right?
>
> Max
>
>
>
> On Tue, Oct 20, 2020 at 3:57 PM Jonas Haase <jhaase at mpe.mpg.de <mailto:jhaase at mpe.mpg.de>> wrote:
>
> Oh dear I never saw Tom’s original mail, appareantly mailing list owners are not automativally subscribers….
>
> Let me anwer now!
>
>> On 20. Oct 2020, at 14:51, Johannes Buchner <jbuchner at mpe.mpg.de <mailto:jbuchner at mpe.mpg.de>> wrote:
>>
>> On 19.10.20 14:58, Tom Dwelly wrote:
>>> Hi Jonas
>>>
>>> I have a few questions about using the MPE SciServer system. I am
>>> sending this to the list in case any of these issues have been
>>> encountered by other users.
>>> Apologies for where I have got the SciServer nomenclature wrong.
>>>
>>>
>>> 1) Is it possible to ssh from an external machine into a container
>>> session running on SciServer compute?
>>
>> This would make SciServer many times more attractive to use for me.
>>
>
> It is not possible to enter a interactive container that way no, however you can use the SciServer Python API to do
> interactive work on your machine using the resources on the SciServer (database, file sharing) and you can also use
> it to launch compute jobs, so running your software inside a new container.
>
> The MPE version of it is here:
> https://github.com/jonashaase/SciScript-Python
>
> I put some small example at the end of the getting started document
> https://datashare.mpcdf.mpg.de/s/1e0CF3yRNcgDL4V
> And there is a example usage file with the code as well.
> Admittedly a lot of it could be made more userfriendly...
>
>> Cheers,
>> Johannes
>>
>>> 2) Is there any way for a user to set up persistent settings (e.g.
>>> .bashrc file, ssh settings etc) that are sourced/used by new terminal
>>> sessions within any compute container?
>>>
>
> You can change .bashrc and everything else the idies user has access to and it should stay the same as long as the
> container is not deleted. However as soon as you start a new container that will all be gone since /home/idies gets
> re-built from the original image.
>
> Only the volumes mounted inside /home/idies/workspace/ are persistent, so you could of course source a setup scrip
> inside your user volume which sets the environment in a way you like.
>
> I am of course happy to adjust the compute images to set up the environment better. I am sure I missed a few things,
> in particular in connection with the x-ray software.
>
>>> 3) Is it possible for users to install standard linux packages (e.g. via
>>> apt) within a container session?
>>>
>
> apt-get not, but conda and pip work.
>
> The reason is that we enforce that the user in the container is ‘idies’ and not root as it most often is in a Docker
> container.
> Its mostly a security issue: We don’t want anyone with root to be able to write onto the mounted disks on the
> underlying system and somehow launch a escalation attack there.
>
>>> 3a) if not, please can you add emacs to the default list of installed
>>> packages in the standard SciServer compute image?
>>>
>
> absolutely, no problem!
>
> @everyone: Please let me know if any software or libraries you need are missing and I’ll install them as quickly as
> possible.
>
>
>>>
>>> 4) Any idea why 'git clone' operations like the following are failing
>>> within a compute container?
>>>
>
> unfortunately not, this should work.
> I will look into it!
>
>
> cheers
> Jonas
>
> —
> Jonas Haase
> Max Planck Institute for Extraterrestrial Physics (MPE)
> Giessenbachstr. 1, 85748 Garching, Germany
> X5 1.3.08
> +49 89 30000 3706
>
> --
> Sciserver-users mailing list
> Sciserver-users at lists.mpe.mpg.de <mailto:Sciserver-users at lists.mpe.mpg.de>
> https://lists.mpe.mpg.de/cgi-bin/mailman/listinfo/sciserver-users
>
>
>
> --
> --------------------------------------------------------------------
> Maximilian Fabricius
> * Max Planck Institute for Extraterrestrial Physics (MPE)
> Giessenbachstrasse, D-85748 Garching, Germany
> eMail: mxhf at mpe.mpg.de <mailto:mxhf at mpe.mpg.de>
> Phone: +49 89 30000 3712
> --------------------------------------------------------------------
More information about the Sciserver-users
mailing list