We've seen already some logs being missed as part of our alert due to falling between this 100ms gap.
Does anyone have any suggestions for me to improve the trigger time so that the gap is closed?
I tried a couple of alternate approaches but it appears that "1m" uses last triggered time as the start of its next scheduled operation. This would explain the drift (though that is speculation).
Instead, I tried a cron based schedule:
"trigger": {
"schedule": {
"cron": "0 */1 * * * ?"
}
},
which, when you look at the trigger_event, seems to force the scheduled time to be on-the-dot:
"trigger_event": {
"type": "schedule",
"triggered_time": "2019-01-28T12:14:00.209Z",
"schedule": {
"scheduled_time": "2019-01-28T12:14:00.000Z"
}
},
Whilst the execution still happens "ms" after the schedule time, the query is now at least scheduling based on the exact minute.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.