feature request: Allow graceful/ quiet hook failures

[goebbe]

Hooks provide means to test if conditions for a backup are satisfied: #33

Examples are hooks that ...

• test if computer is connected to a specific SSID
• test if computer is connected to a metered network
• check if certain timing conditions are met

These examples all define conditions that have to be satisfied for vykar actions (backups ...).
If the conditions are not satisfied, the hook fails and ERROR messages are written to the logs, and the subsequent actions are not executed.

With a high frequency of backups, this can lead to many ERRORs in the log, e.g. in the following situations:

• when travelling (metered networks)
• when not connected to a certain wifi
• when backups are paused at nighttime (by a hook)

While I think it is technically correct to throw an ERROR, in the described cases vykar just works as expected - and the hook failures just lead to the intended behaviour.
The possibly high amount of these ERROR-message could result in a large amount of log entries and ERRORs - and real problems or unexpected ERRORs might stay undiscovered.
Who wants to check if real/ unexpected problems are hiding in a long list of ERROR messages?

Moreover, currently one hook failure can lead to several log-messages. E.g. if my repo-hook fails (vykar-gui) the logs show four messages:

backup: FAILED: hook error ...
prune skipped
compact skipped
check: FAILED: hook error: ...

Perhaps some of these log-messages could be summarized somehow e.g. by "Full backup has been skipped" or similar?

It would be great if hooks would have configurable log feedback that allows to.

• specify own log messages for the case of hook success or hook failure - including the type of message: ERROR, WARN, INFO - that could replace existing log messages

or

• prevent hook - ERROR message from being printed to the logs - on a per-hook basis.

Perhaps there are betters solutions to this type of problem?
Obviously this issue is entirely from a users' perspective.

[goebbe]

This might also be related to: #63

Perhaps a new type of hook could be useful, here?

type	runs when	Failure behaviour
condition	before connection or actions	do not connect, no actions, write INFO to log, reschedule

Purpose: Test if conditions for connections/ actions are met.

If conditions are met:

• Business as usual / connect to repository, do scheduled actions.

If conditions are not met:

• Just print one (configurable) INFO message to the logs. No connections to server, no actions, no info, no additional noise.

Use cases:

• hooks that pause backups
• hooks that check for public metered connections
• hooks that check if connected to specific SSID
• hooks that prevent backups, when laptop battery is critically low
• hock that check for any other imaginable condition (server load ...) that has to be met, before starting actions...

Rational: Failures of these hooks is intended/ planned and useful property - not an ERROR.

Something like this:

hooks:                               
  condition:   >-
    iwgetid | grep -q 'mySSID'
  condition_info: "INFO: not connected to mySSID"