Bug #2332
Differential processing may result in wrong time series data in case of diverging cycle times
| Status: | Resolved | Start date: | 06/29/2015 | |
|---|---|---|---|---|
| Priority: | Normal | Due date: | ||
| Assignee: | J. Wienke | % Done: | 100% | |
| Category: | integration | |||
| Target version: | - |
Description
Assuming a monitored process periodically produces a CPU usage peak. This peak will get noticed by the dbadapter correctly for each produced measurement, but it might get lost in graphite in case the adapter is faster than the configured first retention policy. This can happen because due to timings, the adapter might report something like (0 100% 0 100% 0 100%...) at a higher frequency than the first retention setting and it seems that graphite only takes an arbitrary measurement (and not the average?) in case multiple readings arrive within the specified time frame.
Related issues
History
#1 Updated by J. Wienke almost 9 years ago
- Related to Tasks #2329: Make derivative calculation the default added
#2 Updated by J. Wienke almost 9 years ago
https://graphite.readthedocs.org/en/latest/whisper.html specifies:
Whisper requires that metric updates occur at the same interval as the finest resolution storage archive
This pushes the onus of aggregating values to fit into the finest precision archive to the user rather than the database. It also means that updates are written immediately into the finest precision archive rather than being staged first for aggregation and written later (during a subsequent write operation) as they are in RRD.
Therefore, only the user can ensure this by calling the monitoring programs appropriately. Added to the wiki page: Dashboard_Pipeline_Setup.
#3 Updated by J. Wienke almost 9 years ago
- Status changed from New to Resolved
- % Done changed from 0 to 100
#4 Updated by J. Wienke almost 9 years ago
- Subject changed from Differential processing may result in wrong timer series data in case of diverging cycle times to Differential processing may result in wrong time series data in case of diverging cycle times