Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

install gromacs on travis #41

Closed
2 tasks done
orbeckst opened this issue May 27, 2016 · 14 comments
Closed
2 tasks done

install gromacs on travis #41

orbeckst opened this issue May 27, 2016 · 14 comments
Assignees
Labels
Milestone

Comments

@orbeckst
Copy link
Member

orbeckst commented May 27, 2016

CI on travis is now running (#24) but for integration tests (part of #23) we need a working Gromacs installation.

We would like to test at least:

  • Gromacs 4.6.6
  • Gromacs 5.1.2

See also Becksteinlab/GromacsWrapper#39 – any solution from GromacsWrapper can be applied here.

@ianmkenney
Copy link
Member

We currently have gromacs installed on travis, but it is through Acellera ( mentioned by @dotsdl ) . I am thinking it might just be easier to do a quick and dirty install with the shipped fft libraries. Is this the right way to go?

@dotsdl
Copy link
Contributor

dotsdl commented May 30, 2016

@ianmkenney maybe, but good luck. I think the shortest path to victory here with gromacs 5.x at least is to use the gromacs build installed through conda (from the acellera channel). It should just work, in that with the conda env active the gmx binary should be in the path. Does GromacsWrapper not find it on Travis? I'm too busy to look myself now, but just wanted to chime in.

@orbeckst
Copy link
Member Author

Using acellera conda is fine for the time being. Are there any limitations that we face?

Almost certainly faster than compiling.

We should actually build our own condo gromacs packages… but I am happy to do this once everything else is working.

On 30 May, 2016, at 12:23, Ian Kenney [email protected] wrote:

We currently have gromacs installed on travis, but it is through Acellera ( mentioned by @dotsdl ) . I am thinking it might just be easier to do a quick and dirty install with the shipped fft libraries. Is this the right way to go?


You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or mute the thread.

Oliver Beckstein * [email protected]
skype: orbeckst * [email protected]

@ianmkenney
Copy link
Member

Alright, I was just wondering. I am probably just unfamiliar with Acellera. I will figure out how to get 4.6.6 using it.

@orbeckst
Copy link
Member Author

On 30 May, 2016, at 12:40, Ian Kenney [email protected] wrote:

Alright, I was just wondering. I am probably just unfamiliar with Acellera. I will figure out how to get 4.6.6 using it.

If it’s not in their channel then you cannot get it.

But it DOES exist as a Ubuntu package so just install it through travis. That’s even faster.

You just need some logic to do one or the other and then source the correct GMXRC. But all of it is possible with travis matrix builds. Have a look at the MDAnalysis .travis.yml file.

Oliver Beckstein * [email protected]
skype: orbeckst * [email protected]

@ianmkenney
Copy link
Member

it looks like the version of gromacs on the repos is 4.5.5 instead of 4.6.6. Is there a way to get a different version or do we just use the repo version (I take it that the changes between these versions will be very significant).

@orbeckst
Copy link
Member Author

On 30 May, 2016, at 15:48, Ian Kenney [email protected] wrote:

it looks like the version of gromacs on the repos is 4.5.5 instead of 4.6.6. Is there a way to get a different version or do we just use the repo version (I take it that the changes between these versions will be very significant).

4.5.5 is very old. I don’t think that this will work with MDPOW.

In Ubuntu it is possible to get packages from other repositories. Maybe there’s one available. Search?

Building conda packages looks more and more appealing every minute because then at least we know exactly what we have. The only issue is that you need to build it in a VM with pretty old Linux version (such as old Cent OS). I don’t know the details, though.

@orbeckst
Copy link
Member Author

Maybe there are travis settings that allow you to use a newer Ubuntu releases

      sudo: true
      dist: trusty

Check travis docs.

@orbeckst orbeckst modified the milestone: 0.6 Jun 6, 2016
@ianmkenney
Copy link
Member

@orbeckst it looks like in https://travis-ci.org/Becksteinlab/MDPOW/builds/143638576 we are getting an error with a pip install for recsql. This comes from creating the conda environment defined in environment.yml. We didn't have this issue before did we?

@ianmkenney
Copy link
Member

Funny enough, I was able to pip install in the continuumio/miniconda docker container. The only difference I see is that travis downloaded a tarball while the container download a .whl. Does this different behavior make sense to you?

@ianmkenney
Copy link
Member

Nevermind, it worked on my second push... it downloaded the .whl on travis this time and worked.

@ianmkenney
Copy link
Member

Currently in travis we are creating a single env called mdpow_test. Do we want to have both versions of gromacs installed and used under this env? I assume we are planning to source one and then the other for testing.

@orbeckst
Copy link
Member Author

Separately, not in the same env but as a test matrix.

Oliver Beckstein
email: [email protected]

Am Jul 28, 2016 um 19:40 schrieb Ian Kenney [email protected]:

Currently in travis we are creating a single env called mdpow_test. Do we want to have both versions of gromacs installed and used under this env? I assume we are planning to source one and then the other for testing.


You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or mute the thread.

@orbeckst
Copy link
Member Author

orbeckst commented Aug 2, 2016

Closed with PR #44.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

3 participants