Wednesday 2 November 2016

Re: partman, recipes, sizing, swap, and all the things

On 2 November 2016 at 21:47, Seth Arnold <seth.arnold@canonical.com> wrote:
> On Wed, Nov 02, 2016 at 06:26:07PM +0000, Dimitri John Ledkov wrote:
>> = Questions =
>>
>> Is it sensible to change the semantics of the % specification in partman-auto?
>> Or should e.g. a brand new factor be introduced?
>> Is the introduction of pre-seedable swaplimit sensible?
>> Is 2GB a good generic upper swaplimit?
>> Is hardcoded 2 * 2.5% of disk space a sensible dynamic limit, or
>> should that be preseedable as well?
>
> This proposal feels a little complicated to me. We discussed swap at the
> Athens sprint and I wrote down what I remembered from the discussion:
>
> https://bugs.launchpad.net/ubuntu/+source/debian-installer/+bug/1566129/comments/3
>
> In short:
> - Use swap files
> - Perhaps in a directory, so someone can just dd a new file alongside
> existing ones for additional swap space if locally required
> - One gigabyte is very useful and not too painful
>
> Obviously cloud guests with e.g. 10 gig roots can probably not afford
> an entire gigabyte for swap, so scaling down in this case may still
> be useful.

However, there are filesystems that cannot support swapfiles. For example btrfs.

As far as I understand, swap performance is better on swap formatted
lvm volume, than a swapfile on a filesystem on an lvm volume.

1GB swapfile is indeed expensive on a 10GB rootfs, hence I am hoping
to have a smaller swapfile on really small rootfs.
E.g. 1GB swapfile, but no more than 5% of disk space is simple enough. no?

Are you saying that 2GB is too big, and just do 1GB everywhere will
do, and offer a preseed key to set the size if one wants to change it?
(Or disable out right)

--
Regards,

Dimitri.

--
ubuntu-devel mailing list
ubuntu-devel@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel