Whether to increase the dimension of the metrics #1091
Description
What would you like to be added:
Dimension of the Kmesh‘s metrics.
Why is this needed:
The metrics for kmesh metrics at this stage are:
| Name | Describe |
|---|---|
| kmesh_tcp_workload_connections_opened_total | The total number of TCP connections opened to a workload |
| kmesh_tcp_workload_connections_closed_total | The total number of TCP connections closed to a workload |
| kmesh_tcp_workload_received_bytes_total | The size of the total number of bytes received in response to a workload over a TCP connection |
| kmesh_tcp_workload_sent_bytes_total | The size of the total number of bytes sent in response to a workload over a TCP connection |
| kmesh_tcp_workload_conntections_failed_total | The total number of TCP connections failed to a workload |
And the metrics that kmesh can get directly from the kernel are:
struct tcp_probe_info {
__u32 type;
struct bpf_sock_tuple tuple;
__u32 sent_bytes;
__u32 received_bytes;
__u32 conn_success;
__u32 direction;
__u64 duration; // ns
__u64 close_ns;
__u32 state; /* tcp state */
__u32 protocol;
__u32 srtt_us; /* smoothed round trip time << 3 in usecs */
__u32 rtt_min;
__u32 mss_cache; /* Cached effective mss, not including SACKS */
__u32 total_retrans; /* Total retransmits for entire connection */
__u32 segs_in; /* RFC4898 tcpEStatsPerfSegsIn
* total number of segments in.
*/
__u32 segs_out; /* RFC4898 tcpEStatsPerfSegsOut
* The total number of segments sent.
*/
__u32 lost_out; /* Lost packets */
};But these metrics are connection granularity metrics. Do we need to add these metrics?