docker latest is 9.1 not 9.2

[Abhishek cherath]

Hey everyone,

For some reason the CI build for 9.2 release seems to have failed, so the docker sagemath/sagemath:latest image is of 9.1. Any help I can offer with fixing this?

Thank you,

Abhishek Cherath.

[Dima Pasechnik]

On Thu, Apr 29, 2021 at 9:37 AM Abhishek cherath <abhic...@gmail.com> wrote:

> For some reason the CI build for 9.2 release seems to have failed, so the docker sagemath/sagemath:latest image is of 9.1. Any help I can offer with fixing this?

Thanks for your offer of help.

The docker image is meant to be built by a GitLab CI pipeline, here:
https://gitlab.com/sagemath/dev/sage

But as GitLab own runners are overloaded, we need our own GitLab runner;
there used to be one hosted at Orsay, but I am not sure about its
status, it does not return my calls :-)

Perhaps we should get a GitHub, or some other service, CI pipeline for
this purpose.
(Needless to say, to have a CI pipeline is something to make sure the
container is not containing weird stuff from a unknown to us host, not
only to automate tasks like this)

Please feel free to ask for more details if you feel you can help here.

Best
Dima






>
> Thank you,
> Abhishek Cherath.
>
> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/7ec9c251-2bbe-489b-8ddd-161d57e01cc0n%40googlegroups.com.

[julian...@fsfe.org]

Hi Dima and Abhishek.

On Thursday, April 29, 2021 at 12:44:33 PM UTC+2 dim...@gmail.com wrote:

On Thu, Apr 29, 2021 at 9:37 AM Abhishek cherath <abhic...@gmail.com> wrote:

> For some reason the CI build for 9.2 release seems to have failed, so the docker sagemath/sagemath:latest image is of 9.1. Any help I can offer with fixing this?

Thanks for your offer of help.

The docker image is meant to be built by a GitLab CI pipeline, here:
https://gitlab.com/sagemath/dev/sage

Actually, the one that pushes to docker hub is this one: https://gitlab.com/sagemath/sage

 

But as GitLab own runners are overloaded, we need our own GitLab runner;
there used to be one hosted at Orsay, but I am not sure about its
status, it does not return my calls :-)

The problem is in fact that the GitLab runners have limits that are insufficient to build all of SageMath from scratch (and they are also overloaded and therefore slow to respond.)

 

Perhaps we should get a GitHub, or some other service, CI pipeline for
this purpose.

Afaik the limits on GitHub are also not sufficient to build our docker image in one go. But maybe that has changed by now. The easiest solution would certainly be to get the runners in Orsay working again. Or elsewhere if anybody can volunteer resources. I am happy to assist with the runner setup.

julian

[Dima Pasechnik]

Julian,

Could you please get credentials for an Orsay GitLab runner from Erik?
(Erik offered them to me, but you're certainly more qualified for this task...)

Thanks
Dima

>
> julian

>
> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/b1d0a1a8-3513-4a5d-bc6e-c17606c05d65n%40googlegroups.com.

[E. Madison Bray]

I've restarted the gitlab-runner service. It apparently started
hanging at some point for no apparent reason. Simply kicking it with
a `sudo systemctl restart gitlab-runner` seems to have done the trick.

If anyone would like to have access to the server hosting the
gitlab-runner service please send me an SSH public key.

[Dima Pasechnik]

It's running out of disk space while attempting the build: I see

tee: logs/install.log: No space left on device

in https://gitlab.com/sagemath/sage/-/jobs/1231174714#L1361

>
> If anyone would like to have access to the server hosting the
> gitlab-runner service please send me an SSH public key.
>

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAOTD34b8sTQ5EC6YdawtJXC_KDFKYViWe1ikxmkPem1vbgV6zg%40mail.gmail.com.

[Thierry]

Hi,

On Tue, May 04, 2021 at 04:50:35PM +0100, Dima Pasechnik wrote:
> On Mon, May 3, 2021 at 10:11 AM E. Madison Bray <erik....@gmail.com> wrote:
> >
> > On Thu, Apr 29, 2021 at 2:30 PM julian...@fsfe.org
> > <julian...@fsfe.org> wrote:

[...]

> > I've restarted the gitlab-runner service. It apparently started
> > hanging at some point for no apparent reason. Simply kicking it with
> > a `sudo systemctl restart gitlab-runner` seems to have done the trick.
>
> It's running out of disk space while attempting the build: I see
>
> tee: logs/install.log: No space left on device
>
> in https://gitlab.com/sagemath/sage/-/jobs/1231174714#L1361

Is there a way to manage gitlab runners "à la patchbot" ? I mean with a
simple procedure to allow anyone contribute computational ressources
with no administrative cost, and rootless ?

That would probably be helpful, since many people in the Sage community
will probably be willing to help, there are so many computational
ressources sleeping in many labs (small grids, but also desktops with
lot of cores and RAM which are used to ckeck email and browse the web).

How do running a gitlab runner compares with running a patchbot in terms
of RAM/CPU/Disk/bandwidth cost ?

Ciao,
Thierry

> >
> > If anyone would like to have access to the server hosting the
> > gitlab-runner service please send me an SSH public key.
> >
> > --
> > You received this message because you are subscribed to the Google Groups "sage-devel" group.
> > To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> > To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAOTD34b8sTQ5EC6YdawtJXC_KDFKYViWe1ikxmkPem1vbgV6zg%40mail.gmail.com.
>
> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAAWYfq0JVo5a6aA4NWFJcZ-WKGCs%2B%2BWPoVAiKX8PTVDbWu_-_A%40mail.gmail.com.