Update 22.11 -> 24.11; uptime is changed?

[luddite516]

I am in the process of installing the latest harvest, and will be switching from ZAPI to REST-based data collection.

It appears that the metric 'node_uptime' which existed in 22.11 is not collected by REST. Instead, I see 'uptime' as a label for the 'node_labels' object, and that should work for what we are using it for.

The problem I have is that the 'uptime' label does not appear in the node_labels objects which harvest REST has collected thus far. The templates look good; I have not modified them. But I am not understanding why all the node labels are being collected fine except for uptime?

Example of one of our node_labels instances:

node_labels{bmc_firmware_version="13.10P1", cluster="xxxxxxx", cpu_firmware_release="16.10", datacenter="xxxxxxx", ha_partner="xxxxxxx", healthy="true", instance="localhost:12007", job="xxxxxxx", location="xxxxxxx", max_aggr_size="879609302220800", max_vol_num="2500", max_vol_size="329853488332800", model="FAS8700", node="xxxxxxx", partner_healthy="true", serial="952347003575", site="test2", vendor="NetApp", version="NetApp Release 9.14.1P7: Thu Aug 01 03:07:44 UTC 2024", warnings="There are no failed fans.; There are no failed power supplies.; false"}

[rahulguptajss]

@luddite516 I see that the node_uptime metric is available in version 24.11 for both REST and ZAPI. uptime will not be part of node_labels as it is a metric in template, including it in labels could lead to cardinality issues in Prometheus. I have noticed that uptime is included in the instance_labels within the node template. Since uptime is a metric and not a label, its inclusion is redundant. We will remove it to avoid any confusion through the following pull request: #3404.

Could you please email us the output of the API call api/cluster/nodes?fields=* for this cluster and Rest node template at ng-harvest-files@netapp.com?