8:26 a.m.
Good afternoon everyone,
I am currently using the nice Fedora 30 Vagrant boxes[1] to host some
local dev environment (doing builds) and stumbled upon the fact that the
machines do not provide any swap partition/space (cf attached session
log). My build inside it is thus getting killed when running out of RAM.
I have not found my answer on the list but from some of the searches
I've done[2], it looks like this is a balance between the time spent by
the Fedora cloud team and the users (as to who should do the customization).
Was that the rationale behind this decision to not have swap space into
the images (that users would configure that themselves)?
Thanks for your feedback,
Vincent
[1] attached is my Vagrantfile
[2]
https://lists.fedoraproject.org/archives/list/cloud@lists.fedoraproject.o...
--
www.iot.bzh
9:43 a.m.
On 2/10/20 9:26 AM, Vincent Rubiolo [IoT.bzh] wrote:
Good afternoon everyone,
I am currently using the nice Fedora 30 Vagrant boxes[1] to host some
local dev environment (doing builds) and stumbled upon the fact that the
machines do not provide any swap partition/space (cf attached session
log). My build inside it is thus getting killed when running out of RAM.
I have not found my answer on the list but from some of the searches
I've done[2], it looks like this is a balance between the time spent by
the Fedora cloud team and the users (as to who should do the customization).
Was that the rationale behind this decision to not have swap space into
the images (that users would configure that themselves)?
Thanks for your feedback,
This is mostly because the cloud images don't configure any swap and we keep our
image builds for the vagrant boxes and for the cloud images pretty much the same.
There are two pretty easy options:
- increase the amount of ram for the vagrant box
- configure a second disk in your vagrantfile and then set it up for swap
Dusty
10:25 a.m.
Hi Dusty,
On 2/12/20 4:43 PM, Dusty Mabe wrote:
This is mostly because the cloud images don't configure any swap
and we keep our
image builds for the vagrant boxes and for the cloud images pretty much the same.
Thanks for your quick answer. This makes sense.
Would you have a pointer on which scripts/stack are generating the base
boxes? One way I see would be to remake the base boxes here.
There are two pretty easy options:
- increase the amount of ram for the vagrant box
Yes, this is my current solution, I am doing that in the Vagrantfile.
- configure a second disk in your vagrantfile and then set it up for
swap
Do you have a pointer on how to do that? AFAIS[1] the Vagrant disk
feature is still experimental and not fully documented.
I am using the VirtualBox provider so maybe you were referring to using
the VBoxManage configuration hook[2] or remaking the base box itself[3]
(hence my question above about how the Fedora base boxes are made)?
Thanks again for your help here!
Vincent
[1] https://www.vagrantup.com/docs/disks/usage.html
[2]
https://www.vagrantup.com/docs/virtualbox/configuration.html#vboxmanage-c...
[3] https://www.vagrantup.com/docs/virtualbox/boxes.html#virtual-machine
--
www.iot.bzh
3:53 p.m.
On 2/13/20 11:25 AM, Vincent Rubiolo wrote:
Hi Dusty,
On 2/12/20 4:43 PM, Dusty Mabe wrote:
> This is mostly because the cloud images don't configure any swap and we keep our
> image builds for the vagrant boxes and for the cloud images pretty much the same.
Thanks for your quick answer. This makes sense.
Would you have a pointer on which scripts/stack are generating the base
boxes? One way I see would be to remake the base boxes here.
We generate them using kickstart. Here are the kickstart files (the ones
with *cloud* in the name):
https://pagure.io/fedora-kickstarts/tree/master
>
> There are two pretty easy options:
>
> - increase the amount of ram for the vagrant box
Yes, this is my current solution, I am doing that in the Vagrantfile.
> - configure a second disk in your vagrantfile and then set it up for swap
Do you have a pointer on how to do that? AFAIS[1] the Vagrant disk
feature is still experimental and not fully documented.
I am using the VirtualBox provider so maybe you were referring to using
the VBoxManage configuration hook[2] or remaking the base box itself[3]
(hence my question above about how the Fedora base boxes are made)?
I just add a second disk in my configuration file:
https://github.com/dustymabe/vagrantdirs/blob/430520bcc8fe06430e205366e15...
I use libvirt, so it may be specific to that provider, but I didn't think
it was. You can also try https://www.vagrantup.com/docs/disks/configuration.html
>
> Thanks again for your help here!
>
> Vincent
>
> [1] https://www.vagrantup.com/docs/disks/usage.html
> [2]
>
https://www.vagrantup.com/docs/virtualbox/configuration.html#vboxmanage-c...
> [3] https://www.vagrantup.com/docs/virtualbox/boxes.html#virtual-machine
>
4:02 a.m.
Hi Dusty,
Thanks again for your reply, sorry I had an issue w/ my IMAP folder for
fedora-cloud not being checked automatically, hence the delay (this is
now fixed).
On 2/14/20 10:53 PM, Dusty Mabe wrote:
> Would you have a pointer on which scripts/stack are generating
the base
> boxes? One way I see would be to remake the base boxes here.
We generate them using kickstart. Here are the kickstart files (the ones
with *cloud* in the name):
https://pagure.io/fedora-kickstarts/tree/master
Great pointer, this is exactly what I was looking for, esp
https://pagure.io/fedora-kickstarts/blob/master/f/fedora-cloud-base-vagra...
I use libvirt, so it may be specific to that provider, but I
didn't think
it was. You can also try https://www.vagrantup.com/docs/disks/configuration.html
Ah, indeed. It looks provider-specific here but there seems to be a way
w/ the VirtualBox provider as well, using a VBoxManage[1] hook.
In the meantime, I'll be looking at regenerating the base box using the
KS file you pointed at.
As an aside, I am using the VBox provider because the libvirt one did
not appear to support 'vagrant snapshot' out of the box. I see it should
work w/ sahara instead[2]. Did you try that? Since I am iterating
frequently when debugging provisioning scripts, anything that can speed
up the up/provision/destroy cycle is a benefit. For instance, taking the
base Fedora box, dnf updating, installing needed packages and exporting
the results as a new base box gives a noticeable speed improvement when
bringing up the box another time.
Thanks again,
Vincent
[1]
https://www.vagrantup.com/docs/virtualbox/configuration.html#vboxmanage-c...
[2] https://github.com/vagrant-libvirt/vagrant-libvirt#features
--
www.iot.bzh
1522
days inactive
1530
days old
4 comments
3 participants
participants (3)
-
Dusty Mabe -
Vincent Rubiolo -
Vincent Rubiolo [IoT.bzh]