- We stop services before backup and restart them when backup is completed.
However, if backup fails, we are not restarting the services. With this change,
ensure that stopped services are restarted even if backup process fails.
- Similarly for restore operation.
Tests:
- Backup and restore of an app work.
- Functional tests for matrix-synapse work.
- Run the following two tests without the patch to ensure that the reported bug
is reproducible.
- Make a backup operation fail by raising an exception in the privileged code
that takes backup. Enable matrix-synapse app. Run backup including the
matrix-synapse app. Backup fails and shows an error. The service is
stopped before backup and restarted after backup failure.
- Make a restore operation fail by raising an exception in the privileged code
that does restore. Enable matrix-synapse app. Run backup including the
matrix-synapse app and try to restore it. Restore fails and shows an error. The
service is stopped before restore and restarted after restore failure.
Signed-off-by: Sunil Mohan Adapa <sunil@medhas.org>
Reviewed-by: James Valleroy <jvalleroy@mailbox.org>
Tests:
- DONE: Unit tests work
- DONE: Transmission
- DONE: Enabling/disabling an app with a daemon works: transmission
- DONE: Showing the status of whether the app is enabled with daemon
is-enabled works.
- DONE: A message is shown if app is enabled and service is not running
- DONE: Service is stopped and re-started during backup
- DONE: Adding user to share group during initial setup restarts the service
- Not tested: Enabling/disabling a service with alias works (no such apps)
- DONE: Restarting/try-restarting a service works
- DONE: Masking/unmasking works
- DONE: rsyslog is masked after initial setup
- DONE: systemd-journald is try-restarted during initial setup
- DONE: Avahi, email, security initial setup works
- DONE: Fail2ban is unmasked and enabled
- DONE: Enabling/disabling fail2ban is security app works
- DONE: Enabling/disabling password authentication in SSH works
- ?? Let's encrypt
- Services are try-restarted during certificate setup, obtain, renew
- Not tested: upgrade pagekite from version 1
Signed-off-by: Sunil Mohan Adapa <sunil@medhas.org>
Reviewed-by: James Valleroy <jvalleroy@mailbox.org>
Tests:
- Initial setup works when a new container is created
- When transmission is enabled/disabled, the web configuration for it is
enabled/disabled.
- When radicale is enabled/disabled, the uwsgi configuration for it is
enabled/disabled.
- Sharing web configuration is disabled during backup and re-enabled.
Signed-off-by: Sunil Mohan Adapa <sunil@medhas.org>
Reviewed-by: James Valleroy <jvalleroy@mailbox.org>
Tests performed:
- Schedules are able to store and retrieve comments properly. Information about
schedule backups stored in comments is extracted properly.
- Unit tests run.
Signed-off-by: Sunil Mohan Adapa <sunil@medhas.org>
Reviewed-by: James Valleroy <jvalleroy@mailbox.org>
Each application will be able to implement backup_pre, backup_post, restore_pre
and restore_post hooks that get called before/after backup/restore
appropriately. This is to handle any edge cases that backup manifest mechanism
does not handle.
Reviewed-by: James Valleroy <jvalleroy@mailbox.org>
- Introduce a BackupApp class to store all information about application being
backed up. This cleans up apps lists vs. manifest lists spread out in the
code.
- Introduce ServiceHandler to abstract dealing with services and web
configuration.
- Add enable and disable actions in apache action.
Reviewed-by: James Valleroy <jvalleroy@mailbox.org>
- This serves two purposes. First is to assure user in the interface that backup
of the module is not required. Second is to make sure that if an application
is installed during backup it is also reinstalled during restore process (this
need to be implemented).
- Allow backup test to run independently. Initialize the cfg module so that
load_modules() works.
Reviewed-by: James Valleroy <jvalleroy@mailbox.org>
