We scrape the 32-bit node-runtime counters back into the 64 bit node counters when they overflow.
It should take a good long time for a 32-bit counter to overflow – O(1 seconds’ worth of node runtime) – we do this to reduce the cache footprint of the node-runtime. HTH... Dave From: vpp-dev@lists.fd.io <vpp-dev@lists.fd.io> On Behalf Of nagarajuiit...@gmail.com Sent: Friday, March 27, 2020 1:59 AM To: vpp-dev@lists.fd.io Subject: [vpp-dev] clocks stored in vlib_node_runtime_t #vpp #counters Hi, I see that clib_cpu_time_now() returns u64 type data. And we are updating max clocks spent by a node in vlib_node_runtime_t using these timestamps. Why are we using u32 types in vlib_node_runtime_t? Because of this, it can only hold data within 2 seconds and anything above 32 bit range overflows in our system. Really appreciate your help. Thanks, Nagaraju
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#15896): https://lists.fd.io/g/vpp-dev/message/15896 Mute This Topic: https://lists.fd.io/mt/72581795/21656 Mute #vpp: https://lists.fd.io/mk?hashtag=vpp&subid=1480452 Mute #counters: https://lists.fd.io/mk?hashtag=counters&subid=1480452 Group Owner: vpp-dev+ow...@lists.fd.io Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-