Abnormally high values for some monitoring indicators

PMM2.5.0
image


The monitoring metrics in the above chart are very strange, how can my 4c 8g server be running at 40mil+ questions? The audit logs show that there are not really that many queries, does anyone know why?

Hi @Fan,

All provided charts are based on formulas with rate function. So exporter collected metrics values with a significant increase.

Could you check a pure metric around the issue time in exporter?
e.g. mysql_global_status_created_tmp_files

Did any unusual activity register in service logs around the issue time?

1 Like



1 Like

Hi @Fan,

We have faced similar behavior few times earlier while Prometheus wasn’t replaced by Victoriametric.
VictoriaMetrics treats rate/irate functions in a bit different way that should prevent such issues.

2 Likes

Thank you @Vadim_Yalovets . I’ll try the newer version

1 Like