Feature request: Add (configurable?) number of retries on fail or timeout

[RobinR1]

Once every hour we take an automated backup of the passbolt DB and keys. However this makes passbolt unresponsive or at least very slow during about 1 to 2 minutes.. But since we also run a lot of Ansible plays, with many lookups, it happens quite a lot that we get a timeout error from the lookup plugin at the times such a backup occurs.

So I would like to see the plugin to retry a few times before actually bailing out.

[AnatomicJC]

Hi,

How are you handle the backup of your database ?

If using MariaDB and mysqldump, you should consider mariabackup.

nysqldump freezes the database during backup as mariabackup shouldn't: https://mariadb.com/kb/en/full-backup-and-restore-with-mariabackup/

Cheers,

[AnatomicJC]

Another approach should be to configure a master-slave and perform the backup on the salve instance.

[RobinR1]

If using MariaDB and mysqldump, you should consider mariabackup.

we are using mariadb but the backup is done using the
cake passbolt mysql_export
command. Not sure what method that uses to perform the export.

Another approach should be to configure a master-slave and perform the backup on the salve instance.

that has been added to my wish list now .. :-)

[AnatomicJC]

The cake passbolt mysql_export uses mysqldump, that's why you have a freeze of your environment.

You should consider mariabackup

This tool provides a production-quality, nearly non-blocking method for performing full backups on running systems.

Backup / restore operations: https://mariadb.com/kb/en/full-backup-and-restore-with-mariabackup/

[RobinR1]

Meanwhile I have added a slave database and now backups are performed on the slave side, no longer causing locks/interruptions of the passbolt instance every hour.
However I still get passbolt timeouts every once in a while with the lookup plugin at random times. So an option to retry on timeout or to customize timeout is still feasible.

[Bigouden]

I randomly encounter the same behavior when querying our self-hosted Passbolt. Some requests fail, result in a timeout, and cause the playbook to crash, even when a "retry" option is added to the Ansible tasks. It would be fantastic if the plugin could allow configuring the timeout and/or a retry mechanism.

[Bigouden]

I have implemented the timeout feature in this collection : https://galaxy.ansible.com/ui/repo/published/bigouden/utils/content/lookup/passbolt/

This passbolt plugin supports "PASSBOLT_TIMEOUT" parameter.

It requires the installation of the pip module: passbolt-py (https://pypi.org/project/passbolt-py/), which is a fork I made of py-passbolt that allows setting a timeout on the httpx module : (https://github.com/Bigouden/passbolt-py/blob/main/passbolt/__init__.py#L66)