Thanks, .Rprofile was the culprit. I am still supporting an older
environment which we hope to decommission. After years of using EB, we have
developed some best practices for our site which I was not following,
- Create a special account to own all the software builds, do not build as
yourself
- Build in a container using Docker Run. (or re-image your machine between
builds) Dedicated build machines tend to collect unwanted packages from
unsavory build scripts.
- At a minimum use the toolchain foss-2018b. 2018b only allows single versions
of base libraries to be part of the toolchain. It's important to have
dependency consistency across every package built with a single toolchain.
- Try to use as many published EasyBuild easyconfigs as possible. Resist the
urge to edit and modify easybuilds.
- Contribute back when possible ( --new-pr )
- use PythonPackage and/or RPackage to extend the base R and Python
packages
- Use bundles to create custom versions of base packages.
- Document os dependencies, not every site uses CentOS
- Verify package builds with ldd
John Dey
On 3/30/19, 10:20 AM, "[email protected] on behalf of Lars
Viklund" <[email protected] on behalf of [email protected]>
wrote:
The message `installing to /home/jfdey/R/minqa/libs` in your output
seems a bit strange.
Do you have anything (silly) in your ~/.Renviron or ~/.Rprofile?
If I remember correctly, R is a bit annoying in that it honors flags and
directories specified in those files, which has broken my builds in the
past and are also somewhat documented at
https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_easybuilders_easybuild-2Deasyblocks_issues_867&d=DwIDaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=3TR-iteG1SyRqQ5yubQg-_2KIAToz9bj5dZrRdW36Hc&m=ItDTfh6Nv5MBsyLiyKrxxbdWCcJ40WvM9TYgAcQ98N4&s=Qkgy4PWPbJH4icZClT0AB_7G_PVAU_-NVpgtRRFiixs&e=
.
The `[1] "hello"` part almost looks like the output from shell job
control, but how that'd sneak in, heaven knows.
I set up a 14.04 container and installed R/3.5.3 foss/2018b from a PR,
got past minqa with flying colours.
That one said `installing to
/home/zao/slask/eb-1404/software/R/3.5.3-foss-2018b/lib/R/library/minqa/libs`.
// Lars
On 2019-03-29 23:17, Dey, John F wrote:
> I always install by using --install-latest-eb-release
> After sending this issue I upgrade Python to 2.7.16 on our 14.04 build
machine, and re-installed 3.81. Same results. I have no issue with same setup
on Ubuntu 18.04
>
>
> John Dey
>
> On 3/29/19, 3:05 PM, "Kenneth Hoste" <[email protected]> wrote:
>
> On 29/03/2019 21:24, Dey, John F wrote:
> > I have put my issue into a Gist. This issue is occurring while
building
> > R extensions.
> >
> >
https://urldefense.proofpoint.com/v2/url?u=https-3A__gist.github.com_fizwit_51f41001d5bb58a3525b5f58d84a7606&d=DwIDaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=3TR-iteG1SyRqQ5yubQg-_2KIAToz9bj5dZrRdW36Hc&m=Qx9xK3MqqUx8uKrgUVDD4CinnEVJZB7Y0-pdU3notig&s=b1J8DqAItrUzVFPc4EbLD_Gk3c0tCPVoEglu3JPO6EM&e=
>
> That's positively weird, indeed... Pretty sure it's not EasyBuild
itself
> doing this. Is this a vanilla EasyBuild 3.8.1, or was there some
editing
> done to it?
>
> Anything in your environment that may explain this?
>
> env | grep hello
>
>
> regards,
>
> Kenneth
>
> >
> > *John Dey*
> >
> > *HPC Operations*
> >
> > SciComp Computing
> > *O* _206.667.4308_
> > *M* _360.649.2731_
> >
> > [email protected] <mailto:[email protected]>
> >
> > SciComp Office Hours
> >
> > Wed. 10-noon M4-B102
> >
> >
http://www.fredhutch.org/content/dam/public/email-signatures/3/fred_hutch_logo.png
> > <http://www.fredhutch.org/>
> > Fred Hutchinson Cancer Research Center
> > 1100 Fairview Ave. N., Mail Stop J4-402
> > Seattle, WA 98109
> > *fredhutch.org <http://www.fredhutch.org/>*
> >
>
>
