Re: [PATCH net-next 0/9] ethtool: add uAPI for reading standard stats
From: Ido Schimmel <hidden>
Date: 2021-04-16 15:44:34
On Thu, Apr 15, 2021 at 07:27:43PM -0700, Jakub Kicinski wrote:
Continuing the effort of providing a unified access method to standard stats, and explicitly tying the definitions to the standards this series adds an API for general stats which do no fit into more targeted control APIs. There is nothing clever here, just a netlink API for dumping statistics defined by standards and RFCs which today end up in ethtool -S under infinite variations of names. This series adds basic IEEE stats (for PHY, MAC, Ctrl frames) and RMON stats. AFAICT other RFCs only duplicate the IEEE stats. This series does _not_ add a netlink API to read driver-defined stats. There seems to be little to gain from moving that part to netlink. The netlink message format is very simple, and aims to allow adding stats and groups with no changes to user tooling (which IIUC is expected for ethtool). On user space side we can re-use -S, and make it dump standard stats if --groups are defined.
Jakub, do you have a link for the user space patches? I would like to test it with mlxsw given you already patched it (thanks!).
$ ethtool -S eth0 --groups eth-phy eth-mac eth-ctrl rmon
Given that you have now standardized these stats, do you plan to feed them into some monitoring system? For example, Prometheus has an ethtool exporter [1] and now I see that support is also being added to node_exporter [2] where it really belongs. They obviously mentioned [3] the problem with lack of standardization: "There is also almost no standardization, so if you use multiple network card vendors, you have to examine the data closely to find out what is useful to you and set up your alerts and dashboards accordingly." [1] https://github.com/Showmax/prometheus-ethtool-exporter [2] https://github.com/prometheus/node_exporter/pull/1832 [3] https://tech.showmax.com/2018/11/scraping-ethtool-data-into-prometheus/
Stats for eth0: eth-phy-SymbolErrorDuringCarrier: 0 eth-mac-FramesTransmittedOK: 0 eth-mac-FrameTooLongErrors: 0 eth-ctrl-MACControlFramesTransmitted: 0 eth-ctrl-MACControlFramesReceived: 1 eth-ctrl-UnsupportedOpcodesReceived: 0 rmon-etherStatsUndersizePkts: 0 rmon-etherStatsJabbers: 0 rmon-rx-etherStatsPkts64Octets: 1 rmon-rx-etherStatsPkts128to255Octets: 0 rmon-rx-etherStatsPkts1024toMaxOctets: 1 rmon-tx-etherStatsPkts64Octets: 1 rmon-tx-etherStatsPkts128to255Octets: 0 rmon-tx-etherStatsPkts1024toMaxOctets: 1 v1: Driver support for mlxsw, mlx5 and bnxt included. Compared to the RFC I went ahead with wrapping the stats into a 1:1 nest. Now IDs of stats can start from 0, at a cost of slightly "careful" u64 alignment handling. Jakub Kicinski (9): docs: networking: extend the statistics documentation docs: ethtool: document standard statistics ethtool: add a new command for reading standard stats ethtool: add interface to read standard MAC stats ethtool: add interface to read standard MAC Ctrl stats ethtool: add interface to read RMON stats mlxsw: implement ethtool standard stats bnxt: implement ethtool standard stats mlx5: implement ethtool standard stats Documentation/networking/ethtool-netlink.rst | 74 ++++ Documentation/networking/statistics.rst | 44 +- .../net/ethernet/broadcom/bnxt/bnxt_ethtool.c | 125 ++++++ .../ethernet/mellanox/mlx5/core/en_ethtool.c | 37 ++ .../ethernet/mellanox/mlx5/core/en_stats.c | 142 +++++- .../ethernet/mellanox/mlx5/core/en_stats.h | 10 + .../mellanox/mlxsw/spectrum_ethtool.c | 129 ++++++ include/linux/ethtool.h | 95 ++++ include/uapi/linux/ethtool.h | 10 + include/uapi/linux/ethtool_netlink.h | 137 ++++++ net/ethtool/Makefile | 2 +- net/ethtool/netlink.c | 10 + net/ethtool/netlink.h | 8 + net/ethtool/stats.c | 410 ++++++++++++++++++ net/ethtool/strset.c | 25 ++ 15 files changed, 1248 insertions(+), 10 deletions(-) create mode 100644 net/ethtool/stats.c -- 2.30.2