Investigate why every (fp, timestamp) has two samples with slightly different values

[juliusv]

Example excerpt from a DB dump (fingerprint, timestamp, value):

12602865750567236003-i-4-b 1368268485 1.13312255e+08
12602865750567236003-i-4-b 1368268485 1.13580041e+08
12602865750567236003-i-4-b 1368268545 1.14572348e+08
12602865750567236003-i-4-b 1368268545 1.14811591e+08
12602865750567236003-i-4-b 1368268605 1.15656432e+08
12602865750567236003-i-4-b 1368268605 1.15890474e+08
12602865750567236003-i-4-b 1368268665 1.1688117e+08
12602865750567236003-i-4-b 1368268665 1.17082066e+08
12602865750567236003-i-4-b 1368268725 1.18143149e+08
12602865750567236003-i-4-b 1368268725 1.18359716e+08
12602865750567236003-i-4-b 1368268785 1.19380636e+08
12602865750567236003-i-4-b 1368268785 1.19668582e+08
12602865750567236003-i-4-b 1368268845 1.20465585e+08
12602865750567236003-i-4-b 1368268845 1.20714068e+08
12602865750567236003-i-4-b 1368268905 1.21626873e+08
12602865750567236003-i-4-b 1368268905 1.21865351e+08
12602865750567236003-i-4-b 1368268965 1.22761552e+08

This is before compaction, mind you.

[juliusv]

/cc @matttproud

[juliusv]

Solved. This happened because I scraped two active/passive HAProxy-exporters and their differentiating instance labels got dropped in favor of the ones specified in the metrics.json (where instance=backend). Time to implement the behavior where we instead set exporter_instance etc. when the exported metrics contains these special labels.

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.