You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Divide the disk free bytes values by 4096 and a pattern emerges. Note that in this series the first negative value of each spike series is -previous_value - 20.
I got back to investigating this further and it seems the input from telegraf to influxdb is correct (no negative values), yet there are negative values returned from influxdb when querying the raw data. Could be related to influxdata/influxdb#6643
Closing this ticket and continuing on the influxdb side.
Bug report
Series of negative free disk space spikes observed on some Linux nodes.
System info:
Expected behavior:
Correct disk free values expected.
Actual behavior:
Sample of raw data from InfluxDB:
Additional info:
Divide the disk free bytes values by 4096 and a pattern emerges. Note that in this series the first negative value of each spike series is
-previous_value - 20
.I have not yet verified the actual input going from Telegraf to InfluxDB to confirm that this is not an InfluxDB issue.
I tried reviewing the relevant code, but wasn't able to spot any obvious place where the value could turn negative:
telegraf/internal/models/makemetric.go
Line 114 in b00ad65
telegraf/plugins/inputs/system/disk.go
Line 66 in 6648c10
telegraf/plugins/inputs/system/ps.go
Line 86 in 6648c10
The text was updated successfully, but these errors were encountered: