e.g.
"It's unfortunately a global 'ok' state. If all checks passed, .ok() is called. When there are 5 checks in the script that could have called .warning(), .ok() would have no way to know which check to report on.
Similarly, monitorlib has no way to report multiple warnings/failures per check instance.
Say, check_http fails 3 URLs, it only reports the first one (to avoid emailing the warn state every 60 seconds).
This is a future enhancement - we should really track states whereby multiple instance/last_values can be tracked for each check.
-Charlie
On Feb 11, 2013, at 12:40 PM, Jos Boumans jos@krux.com wrote:
Would it be useful to have the new state show up in the message AFTER
an alert (or even in general)? Last error was '75.95ms is over 75ms'.
I now wonder, are we at 74.95, or at 30ms? Are we really ok?
On 11 Feb 2013, at 12:37, collectd@krux.com
wrote:
{'host': 'cc001', 'message': 'all is well', 'severity': 'okay', 'plugin': 'response_times.py'}
"
e.g.
"It's unfortunately a global 'ok' state. If all checks passed, .ok() is called. When there are 5 checks in the script that could have called .warning(), .ok() would have no way to know which check to report on.
Similarly, monitorlib has no way to report multiple warnings/failures per check instance.
Say, check_http fails 3 URLs, it only reports the first one (to avoid emailing the warn state every 60 seconds).
This is a future enhancement - we should really track states whereby multiple instance/last_values can be tracked for each check.
-Charlie
On Feb 11, 2013, at 12:40 PM, Jos Boumans jos@krux.com wrote:
Would it be useful to have the new state show up in the message AFTER
an alert (or even in general)? Last error was '75.95ms is over 75ms'.
I now wonder, are we at 74.95, or at 30ms? Are we really ok?
On 11 Feb 2013, at 12:37, collectd@krux.com
wrote:
{'host': 'cc001', 'message': 'all is well', 'severity': 'okay', 'plugin': 'response_times.py'}
"