Currently we write points to output one at a time.
It'd be far better to batch into a single write.
assigned to @btasker
Points are now collected into a list and a single write is performed (per output).
We're actually doing it the more inefficient way round at the moment though - we iterate over outputs and have to construct a points buffer for each.
Iterating over outputs first was the more efficient approach the way we were doing it before, but really it now needs turning on it's head.
mentioned in commit github-mirror/tplink_to_influxdb@67aa0fd3b003c2261191e3ab30cdc7980313aad8
Commit: github-mirror/tplink_to_influxdb@67aa0fd3b003c2261191e3ab30cdc7980313aad8 Author: B Tasker Date: 2022-09-06T14:44:00.000+01:00
Batch points into a single write for utilities/tp-link-to-influxdb#1
Points will be pushed into a list and then a single write made.
mentioned in commit github-mirror/tplink_to_influxdb@36e2640f0a73345b0cf15ca46b3fe71737bdf279
Commit: github-mirror/tplink_to_influxdb@36e2640f0a73345b0cf15ca46b3fe71737bdf279 Author: B Tasker Date: 2022-09-06T14:55:47.000+01:00
Change order of processing so that we only need to construct a list of datapoints once, rather than once per output (utilities/tp-link-to-influxdb#1)
Activity
03-Sep-22 23:08
assigned to @btasker
06-Sep-22 13:51
Points are now collected into a list and a single write is performed (per output).
We're actually doing it the more inefficient way round at the moment though - we iterate over outputs and have to construct a points buffer for each.
Iterating over outputs first was the more efficient approach the way we were doing it before, but really it now needs turning on it's head.
06-Sep-22 14:00
mentioned in commit github-mirror/tplink_to_influxdb@67aa0fd3b003c2261191e3ab30cdc7980313aad8
Message
Batch points into a single write for utilities/tp-link-to-influxdb#1
Points will be pushed into a list and then a single write made.
06-Sep-22 14:00
mentioned in commit github-mirror/tplink_to_influxdb@36e2640f0a73345b0cf15ca46b3fe71737bdf279
Message
Change order of processing so that we only need to construct a list of datapoints once, rather than once per output (utilities/tp-link-to-influxdb#1)