Hi!

I see you've set version 0.4 in meta/main.yml - could I trouble you to create a git tag and push that tag to github?

Should be as simple as 'git tag 0.4' and then 'git push --tags'. I hope :)

Thanks!

Could semantic versioning be used?

Repo link to Ansible Galaxy Role https://galaxy.ansible.com/detail#/role/1262 is a dead link. As in it doesn't load anything at the moment.

I just set up swap using the default which should be 512MB

---
swapfile_location: /swapfile
swapfile_size: 512MB
swapfile_swappiness: False
swapfile_vfs_cache_pressure: False
swapfile_use_dd: False

But when I check my Digital Ocean Ubuntu 17.4. 64 bits droplet I see 488 kilobits:

root@larastudio:~# swapon -s
Filename				Type		Size	Used	Priority
/swapfile                              	file    	499996	0	-1

And when I run ls -lh /swapfile I get less than 512MB or 489:

root@larastudio:~# ls -lh /swapfile
-rw------- 1 root root 489M Jun 14 09:36 /swapfile

Any ideas why?

It seems like this role is tearing down and setting up the swapfile every time I run my playbook. I am pretty sure that it left the swapfile alone if it was already appropriate in the last version. Is it possible to fix the check as to whether it needs to do anything or not?

E.g. https://github.com/geerlingguy/ansible-role-mysql/blob/master/.travis.yml

Hi

Thanks for a great ansible role :)

I was wondering if you'd consider changing the behaviour of the swappiness / cache pressure?

I think the most normal use of Ansible is on servers rather than on desktops - and in that case the default swappiness/cache pressure values are probably most appropriate.

I've come up with a few options/ideas here:

Split Roles

Split the function of configuring swappiness and other values into a new and separate role like kamaln7.swapparams or something similar.

Default to keeping existing config

If no variable is set, then no change is made - and the system keeps the OS defaults.

This way if someone includes a different role that also manages the values, there'll be no conflict. This is likely fiddly to do at an Ansible level, since it might have to inspect the existing /proc values

Use the Linux defaults

If you default the values to the standard linux values of 60/100 it'd at least mean no unexpected changes in most cases.

Document

Just clearly mark that the values will change in an example somewhere :) This is probably easiest!

Thanks again!

Oskar

@kamaln7 I added become: true and become_method: sudo to these actions for my use case. Any inputs for making it generic? Will be happy to submit a PR.

Either a bug or a warning, {{ swapfile_size }} having the units in the example doesn't work if you fall-back to dd, as the corresponding "count" parameter for dd should just be an integer. If you give it units, it will obediently convert them back to bytes first.

Worked example with

swapfile_size=512MB
swapfile_use_dd=False
--
fallocate -l {{ swapfile_size }} {{ swapfile_location }} creates={{ swapfile_location }}

will create a 512MB swapfile, but

swapfile_size=512MB
swapfile_use_dd=True
--
dd if=/dev/zero of={{ swapfile_location }} bs=1M count={{ swapfile_size }} creates={{ swapfile_location }}

will create a 512000000MB or 512 terabyte swap file, which for my purposes is a little excessive.

On an ext3 filesystem, fallocate fails with:

stderr: fallocate: /swapfile: fallocate failed: Operation not supported

from the man page:

      As  of the Linux Kernel v2.6.31, the fallocate system call is supported
       by the btrfs, ext4, ocfs2, and xfs filesystems.

...either an error message when the filesystem is unsupported or alternate allocation type would be appreciated.

A more reliable alternative is to use dd instead of fallocate:

sudo dd if=/dev/zero of=/mnt/512MiB.swap bs=1024 count=524288
sudo chmod 600 /mnt/512MiB.swap

see: https://help.ubuntu.com/community/SwapFaq#How_do_I_add_more_swap.3F