Issue:
f5_network_interface_bits_in_total does not seem to be correct when compared to the other metrics for pool and node.
AST version is v0.9.6.
Customer issue:
The customer is pulling the Cisco metrics for the ports and wants to match the data volume to the big-ip interface. This does not work, the values are drastically different but it does match using the pool or node.
Explanation:
On a simple big-ip config pulling in the total pool, node and external interface data the total bits/bytes are not adding up to the same value after the recommended "/8", the graphing trends look to be very similar patterns but the interface is a lot bigger in values.
Expectation was they should match interface in total data to the pools in as all traffic is coming in.
Metrics being used are:
- f5_node_bytes_in_total
- f5_pool_bytes_in_total
- f5_network_interface_bits_in_total
The interface data if divided by 4 matches, not sure why it should be divides by 8 to move from bits to bytes per the prometheus and demo dashboards.
Reproducing:
This was reproducible on the F5 internal UDF AST lab regarding the interface data.
Divide by 4 on interface - similar graphs:
Divide by 8 on interface - interface drops, not possible it should match the pool/node in:
Issue:
f5_network_interface_bits_in_total does not seem to be correct when compared to the other metrics for pool and node.
AST version is v0.9.6.
Customer issue:
The customer is pulling the Cisco metrics for the ports and wants to match the data volume to the big-ip interface. This does not work, the values are drastically different but it does match using the pool or node.
Explanation:
On a simple big-ip config pulling in the total pool, node and external interface data the total bits/bytes are not adding up to the same value after the recommended "/8", the graphing trends look to be very similar patterns but the interface is a lot bigger in values.
Expectation was they should match interface in total data to the pools in as all traffic is coming in.
Metrics being used are:
The interface data if divided by 4 matches, not sure why it should be divides by 8 to move from bits to bytes per the prometheus and demo dashboards.
Reproducing:
This was reproducible on the F5 internal UDF AST lab regarding the interface data.
Divide by 4 on interface - similar graphs:
Divide by 8 on interface - interface drops, not possible it should match the pool/node in:
