ansible-collection-hardening
ansible-collection-hardening copied to clipboard
rework filesystem hardening
- removed a lot duplicated code by using a loop
- added new hardening options for /tmp
- added new options "passno" and "dump" for every filesystem. currently ansible changed that values to 0 for every fs new default depends on fstype, can be overwritten in config
- removed default fstype in config the type will now be autodetected, can be overwritten in config
- mount src setting is now optional the source will now be autodetected, can be overwritten in config
- it will be now checked, if it is really a mount
- changed fs reload to handler
- removed check os_auditd_enabled on /var/log/audit
Notes:
- The playbook will no longer fail, if a folder does not exists
- It will now skip mount points, if they are not present on the target system. If you have many different configured servers this is very useful
- The folders are still be hardened regardless if *_enabled is used or not
- Was there a reason why /var/log/audit was only hardened, when the os_auditd_enabled variable was set to true?
i think this would supersede #523
Yes, I was too slow :) Thanks! #523 is deprecated
removed a lot duplicated code by using a loop
added new hardening options for /tmp
added new options "passno" and "dump" for every filesystem. currently ansible changed that values to 0 for every fs new default depends on fstype, can be overwritten in config
removed default fstype in config the type will now be autodetected, can be overwritten in config
mount src setting is now optional the source will now be autodetected, can be overwritten in config
it will be now checked, if it is really a mount
changed fs reload to handler
removed check os_auditd_enabled on /var/log/audit
Notes:
* The playbook will no longer fail, if a folder does not exists * It will now skip mount points, if they are not present on the target system. If you have many different configured servers this is very useful * The folders are still be hardened regardless if *_enabled is used or not * Was there a reason why /var/log/audit was only hardened, when the os_auditd_enabled variable was set to true?
Hi @divialth,
yes there was a reason for me, because /var/log/audit only exists on RHEL derivates. Debian based does not have this folder. And some users do not want to harden /var/log/audit so we set the default to false.
Would you prefer to query a fact for an specific operatingsystems to set this to true?
yes there was a reason for me, because
/var/log/auditonly exists on RHEL derivates. Debian based does not have this folder. And some users do not want to harden/var/log/auditso we set the default tofalse.Would you prefer to query a fact for an specific operatingsystems to set this to
true?
This should be no longer a problem. The first task in minimize_access_fs.yml only runs, when it is real mount point (or a whitelisted special device like /run, /dev...)
The hardening of the permissions on the directory also does now only run, if it is already existing.
I also did not changed any of the _enabled defaults.
The only thing i removed was the boolean check for os_auditd_enabled.
I think this will work for us. A "go" on my side :+1:
please consider my suggestions as nitpicking.
Aside from my comments I want to say I absolutely like what you did here!
I have implemented your other naming suggestions and also did a rebase to resolve the merge conflicts. The current failing CI checks seems to be unrelated. Please correct me if I am wrong.