I have promql query
process_resident_memory_bytes{job="integrations/kubernetes/kube-apiserver", cluster=~"my-cluster"}
Which gives following data in grafana portal
| Timestamp | Value |
|---|---|
| 2023-09-23 17:30:00 | 1097273344 |
| 2023-09-23 17:30:30 | 1097273344 |
| 2023-09-23 17:31:00 | 1098895360 |
| 2023-09-23 17:31:30 | 1098895360 |
| 2023-09-23 17:32:00 | 1082576896 |
| 2023-09-23 17:32:30 | 1082576896 |
| 2023-09-23 17:33:00 | 1116635136 |
| 2023-09-23 17:33:30 | 1116635136 |
| 2023-09-23 17:34:00 | 1110745088 |
| 2023-09-23 17:34:30 | 1110745088 |
| 2023-09-23 17:35:00 | 1072701440 |
| 2023-09-23 17:35:30 | 1072701440 |
| 2023-09-23 17:36:00 | 1056124928 |
| 2023-09-23 17:36:30 | 1056124928 |
| 2023-09-23 17:37:00 | 1193168896 |
| 2023-09-23 17:37:30 | 1193168896 |
| 2023-09-23 17:38:00 | 1028886528 |
| 2023-09-23 17:38:30 | 1028886528 |
| 2023-09-23 17:39:00 | 1080430592 |
| 2023-09-23 17:39:30 | 1080430592 |
| 2023-09-23 17:40:00 | 1087283200 |
| 2023-09-23 17:40:30 | 1087283200 |
| 2023-09-23 17:41:00 | 1064062976 |
| 2023-09-23 17:41:30 | 1064062976 |
| 2023-09-23 17:42:00 | 1176289280 |
| 2023-09-23 17:42:30 | 1176289280 |
| 2023-09-23 17:43:00 | 1094549504 |
| 2023-09-23 17:43:30 | 1094549504 |
| 2023-09-23 17:44:00 | 1059459072 |
| 2023-09-23 17:44:30 | 1059459072 |
| 2023-09-23 17:45:00 | 1064747008 |
| 2023-09-23 17:45:30 | 1064747008 |
| 2023-09-23 17:46:00 | 1052381184 |
| 2023-09-23 17:46:30 | 1052381184 |
| 2023-09-23 17:47:00 | 1140494336 |
| 2023-09-23 17:47:30 | 1140494336 |
| 2023-09-23 17:48:00 | 1073520640 |
| 2023-09-23 17:48:30 | 1073520640 |
| 2023-09-23 17:49:00 | 1080832000 |
| 2023-09-23 17:49:30 | 1080832000 |
| 2023-09-23 17:50:00 | 1082146816 |
| 2023-09-23 17:50:30 | 1082146816 |
| 2023-09-23 17:51:00 | 1087004672 |
| 2023-09-23 17:51:30 | 1087004672 |
| 2023-09-23 17:52:00 | 1176756224 |
| 2023-09-23 17:52:30 | 1176756224 |
| 2023-09-23 17:53:00 | 1098498048 |
| 2023-09-23 17:53:30 | 1098498048 |
| 2023-09-23 17:54:00 | 1103314944 |
| 2023-09-23 17:54:30 | 1103314944 |
| 2023-09-23 17:55:00 | 1083441152 |
| 2023-09-23 17:55:30 | 1083441152 |
| 2023-09-23 17:56:00 | 1081024512 |
| 2023-09-23 17:56:30 | 1081024512 |
| 2023-09-23 17:57:00 | 1118924800 |
| 2023-09-23 17:57:30 | 1118924800 |
| 2023-09-23 17:58:00 | 1038856192 |
| 2023-09-23 17:58:30 | 1038856192 |
| 2023-09-23 17:59:00 | 1082642432 |
| 2023-09-23 17:59:30 | 1082642432 |
| 2023-09-23 18:00:00 | 1083723776 |
After applying the following query, I have obtained the following data.
increase(process_resident_memory_bytes{job="integrations/kubernetes/kube-apiserver", cluster=~"my-cluster"}[10m])
| Timestamp | Value |
|---|---|
| 2023-09-23 17:50:00 | 4993515520 |
| 2023-09-23 17:50:30 | 4993515520 |
| 2023-09-23 17:51:00 | 4874217244.444445 |
| 2023-09-23 17:51:30 | 4874217244.444445 |
| 2023-09-23 17:52:00 | 3757775075.555556 |
| 2023-09-23 17:52:30 | 3757775075.555556 |
| 2023-09-23 17:53:00 | 3801151715.555556 |
| 2023-09-23 17:53:30 | 3801151715.555556 |
| 2023-09-23 17:54:00 | 3800628337.777778 |
| 2023-09-23 17:54:30 | 3800628337.777778 |
| 2023-09-23 17:55:00 | 3835139413.3333335 |
| 2023-09-23 17:55:30 | 3835139413.3333335 |
| 2023-09-23 17:56:00 | 4938374257.777778 |
| 2023-09-23 17:56:30 | 4938374257.777778 |
| 2023-09-23 17:57:00 | 3787684977.777778 |
| 2023-09-23 17:57:30 | 3787684977.777778 |
| 2023-09-23 17:58:00 | 4933845902.222222 |
| 2023-09-23 17:58:30 | 4933845902.222222 |
| 2023-09-23 17:59:00 | 4981036373.333334 |
| 2023-09-23 17:59:30 | 4981036373.333334 |
| 2023-09-23 18:00:00 | 4976840248.888889 |
My understanding increase function formula is value[currentvalue] - value[current-10min].
Example to calculate the increase value at particular timestamp:2023-09-23 18:00:00
1083723776 - 1082146816(i.e value at 2023-09-23 17:50:00) = 1576960
The output I obtained in the second table differs at "2023-09-23 18:00:00." Please guide me on what might be missing when I try to calculate the increase function, or correct me if my understanding is incorrect.
Thanks in Advance.
No, increase is more complex then a simple subtraction. It automatically accounts for counter resets, that are present in data, and thus should only be used for counters.
Your metric being called process_resident_memory_bytes is not a counter. If you want to simply get change in value over last 10 minutes use either delta function, or if you are concerned about non-integer results something like
process_resident_memory_bytes{job="integrations/kubernetes/kube-apiserver", cluster=~"my-cluster"} - process_resident_memory_bytes offset 10m