intel_dlb

Intel® Dynamic Load Balancer Input Plugin

This plugin collects metrics exposed by applications built with the Data Plane Development Kit, an extensive set of open source libraries designed for accelerating packet processing workloads, plugin is also using bifurcated driver. More specifically it's targeted for applications using Intel DLB as eventdev devices accessed via bifurcated driver (allowing access from kernel and user-space).

⭐ Telegraf v1.25.0 🏷️ applications 💻 linux

Requirements

• Linux kernel 5.12+
• DLB >= v7.4
• DPDK >= 20.11.3

[!NOTE] It may happen that sysfs entries or the socket telemetry interface exposed by DPDK-based app will require root access. This means that either access permissions have to be adjusted for sysfs / socket telemetry interface to allow Telegraf to access it, or Telegraf should run with root privileges.

Global configuration options

In addition to the plugin-specific configuration settings, plugins support additional global and plugin configuration settings. These settings are used to modify metrics, tags, and field or create aliases and configure ordering, etc. See the CONFIGURATION.md for more details.

Configuration

## Reads metrics from DPDK using v2 telemetry interface.
## This plugin ONLY supports Linux
[[inputs.intel_dlb]]
  ## Path to DPDK telemetry socket.
  # socket_path = "/var/run/dpdk/rte/dpdk_telemetry.v2"

  ## Default eventdev command list, it gathers metrics from socket by given commands.
  ## Supported options:
  ##   "/eventdev/dev_xstats", "/eventdev/port_xstats",
  ##   "/eventdev/queue_xstats", "/eventdev/queue_links"
  # eventdev_commands = ["/eventdev/dev_xstats", "/eventdev/port_xstats", "/eventdev/queue_xstats", "/eventdev/queue_links"]

  ## Detect DLB devices based on device id.
  ## Currently, only supported and tested device id is `0x2710`.
  ## Configuration added to support forward compatibility.
  # dlb_device_types = ["0x2710"]

  ## Specifies plugin behavior regarding unreachable socket (which might not have been initialized yet).
  ## Available choices:
  ##   - error: Telegraf will return an error on startup if socket is unreachable
  ##   - ignore: Telegraf will ignore error regarding unreachable socket on both startup and gather
  # unreachable_socket_behavior = "error"

Default configuration allows getting metrics for all metrics reported via /eventdev/ command:

• /eventdev/dev_xstats
• /eventdev/port_xstats
• /eventdev/queue_xstats
• /eventdev/queue_links

Metrics

There are two sources of metrics:

• DPDK-based app for detailed eventdev metrics per device, per port and per queue
• Sysfs entries from kernel driver for RAS metrics

Example Output

intel_dlb,command=/eventdev/dev_xstats\,0,host=controller1 dev_dir_pool_size=0i,dev_inflight_events=8192i,dev_ldb_pool_size=8192i,dev_nb_events_limit=8192i,dev_pool_size=0i,dev_rx_drop=0i,dev_rx_interrupt_wait=0i,dev_rx_ok=463126660i,dev_rx_umonitor_umwait=0i,dev_total_polls=78422946i,dev_tx_nospc_dir_hw_credits=0i,dev_tx_nospc_hw_credits=584614i,dev_tx_nospc_inflight_credits=0i,dev_tx_nospc_inflight_max=0i,dev_tx_nospc_ldb_hw_credits=584614i,dev_tx_nospc_new_event_limit=59331982i,dev_tx_ok=694694059i,dev_zero_polls=29667908i 1641996791000000000
intel_dlb,command=/eventdev/queue_links\,0\,1,host=controller1 qid_0=128i,qid_1=128i 1641996791000000000
intel_dlb_ras,device=pci0000:6d,host=controller1,metric_file=aer_dev_correctable BadDLLP=0i,BadTLP=0i,CorrIntErr=0i,HeaderOF=0i,NonFatalErr=0i,Rollover=0i,RxErr=0i,TOTAL_ERR_COR=0i,Timeout=0i 1641996791000000000
intel_dlb_ras,device=pci0000:6d,host=controller1,metric_file=aer_dev_fatal ACSViol=0i,AtomicOpBlocked=0i,BlockedTLP=0i,CmpltAbrt=0i,CmpltTO=0i,DLP=0i,ECRC=0i,FCP=0i,MalfTLP=0i,PoisonTLPBlocked=0i,RxOF=0i,SDES=0i,TLP=0i,TLPBlockedErr=0i,TOTAL_ERR_FATAL=0i,UncorrIntErr=0i,Undefined=0i,UnsupReq=0i,UnxCmplt=0i 1641996791000000000