opareport(8)

📦 version 10.14.4.0.20, in centos-stream 9, debian bookworm, debian trixie, opensuse-leap 16.0, ubuntu jammy, ubuntu noble, ubuntu questing, ubuntu resolute, ubuntu stonking

💖 opareport — the IFSFFCLIRG (Man Page) glow-up ✨

NAME 🌸

opareport

Bestie. Okay so. This is the tool for fabric analysis and reporting — it's giving main character energy for your whole fabric. It's (All) flavors of useful, and it's got powerful fabric analysis and reporting capabilities, no cap.

🚨 Real talk though: you must run this on a host that's connected to the Intel(R) Omni-Path Fabric, and you must have the Intel(R) Omni-Path Fabric Suite FastFabric Toolset installed. No toolset, no report, periodt.

Syntax ✨

opareport  [-v][-q] [-h  hfi] [-p  port] [--timeout] [-o  report]
[-d  detail] [-P|-H] [-N] [-x] [-X  snapshot_input] [-T  topology_input]
[-s] [-r] [-V] [-i seconds] [-b  date_time] [-e  date_time] [-C] [-a]
[-m] [-M] [-A] [-c  file] [-L] [-F  point] [-S  point] [-D  point] [-Q]

Options 💅

  • --help — spills the full help text. Like, the whole thing. Full tea.
  • -v/--verbose — returns verbose output. Extra detail, extra yapping, we love her for it.
  • -q/--quiet — disables progress reports. She's shy now. No running commentary.
  • -h/--hfi hfi — specifies the HFI, numbered 1..n. Using 0 specifies that the -p port port is a system-wide port number. Default is 0.
  • -p/--port port — specifies the port, numbered 1..n. Using 0 specifies the first active port. Default is 0.
  • --timeout — specifies the timeout (that's the wait time for a response, iykyk) in ms. Default is 1000 ms.
  • -o/--output report — specifies the report type for output. Peep the Report Types section for the deets.
  • -d/--detail level — specifies the level of detail 0-n for output. Default is 2.
  • -P/--persist — only includes data persistent across reboots.
  • -H/--hard — only includes permanent hardware data.
  • -N/--noname — omits node and IOC names. Anonymous era. 🙈
  • -x/--xml — produces output in XML.
  • -X/--infile snapshot_input — generates a report using the data in the snapshot_input file. That snapshot_input must have been generated during a previous -o snapshot run. When you use this, the -s, -i, -C, and -a options are ignored (they're just not invited to the party). '-' may be used as the snapshot_input to specify stdin.
  • -T/--topology topology_input — uses the topology_input file to augment and verify fabric information. When you use it, various reports can be augmented with info that isn't available electronically — like cable labels, bestie, the physical receipts. '-' may be used to specify stdin.
  • -s/--stats — gets performance statistics for all ports.
  • -i/--interval seconds — obtains performance statistics over interval seconds. It clears all statistics, waits interval seconds, then generates the report. Implies -s option.
  • -b/--begin date_time — obtains performance stats beginning at date_time. Implies -s option.
    • If -b/--begin and -e/--end are specified, the performance stats are provided over the specified interval.
    • If only -b/--begin is specified, the absolute value of the performance stats at the specified time is provided.

date_time may be a time entered as HH:MM[:SS] or date as mm/dd/YYYY, dd.mm.YYYY, YYYY-mm-dd or date followed by time; for example, "2016-07-04 14:40". Relative times are taken as "x [second|minute|hour|day](s) ago.

  • -e/--end date_time — obtains performance stats ending at date_time. Implies -s option.
    • If -b/--begin and -e/--end are specified, the performance stats are provided over the specified interval.
    • If only -e/--end is specified, the absolute value of the performance stats at the specified time is provided.

date_time may be a time entered as HH:MM[:SS] or date as mm/dd/YYYY, dd.mm.YYYY, YYYY-mm-dd or date followed by time; for example, "2016-07-04 14:40". Relative times are taken as "x [second|minute|hour|day](s) ago.

  • -C/--clear — clears performance statistics for all ports. Heads up: only statistics with error thresholds are cleared. And the clear happens after generating the report. Order matters, babe. 🎀
  • -a/--clearall — clears all performance statistics for all ports. Everything. Gone. Iconic but also terrifying.
  • -m/--smadirect — accesses fabric information directly from SMA.
  • -M/--pmadirect — accesses performance statistics using direct PMA.
  • -A/--allports — gets PortInfo for down switch ports. Uses direct SMA to get this data. If used with -M, also gets PMA stats for down switch ports.
  • -c/--config file — specifies the error thresholds configuration file. Default is the /etc/opa/opamon.conf file.
  • -L/--limit — limits operation to the exact specified focus with -F for the port error counters check (-o errors) and the port counters clear (-C or -i). Normally, the neighbor of each selected port is also checked/cleared. Does not affect other reports. She stays in her lane.
  • -F/--focus point — specifies the focus area for the report.

Limits output to reflect a subsection of the fabric. Heads up though: it may not work with all reports. (For example, route, mcgroups, and the verify* reports may ignore the option or not generate useful results.) So like… she's picky. 😭

  • -S/--src point — specifies the source for the trace route. Default is local port.
  • -D/--dest point — specifies the destination for the trace route.
  • -Q/--quietfocus — excludes the focus description from the report.

🦋 The -h and -p combo menu 🦋

Okay so -h and -p options permit a variety of selections. Pick your fighter:

  • -h 0 — First active port in system (Default).
  • -h 0 -p 0 — First active port in system.
  • -h x — First active port on HFI x.
  • -h x -p 0 — First active port on HFI x.
  • -h 0 -p y — Port y within system (no matter which ports are active).
  • -h x -p y — HFI x, port y.

Snapshot-Specific Options 📸

  • -r/--routes — gets routing tables for all switches.
  • -V/--vltables — gets the P-Key tables for all nodes and the QoS VL-related tables for all ports. Two for one, ate. 👑

🌈✨ Report Types ✨🌈

okay bestie, so these are all the report types you can ask for. each one is a different flavor of tea, and you literally just pick the one you need. let's get into it:

  • comps — a summary of every single system and SM in the fabric. it's giving the whole squad. 💖
  • brcomps — a brief summary of all systems and SMs in the fabric. the TL;DR version, no cap. ✨
  • nodes — a summary of all node types and SMs in the fabric.
  • brnodes — a brief summary of all node types and SMs in the fabric.
  • ious — a summary of all IO units in the fabric.
  • lids — a summary of all LIDs in the fabric.
  • linkinfo — a summary of all links with LIDs in the fabric.
  • links — a summary of all links.
  • extlinks — a summary of links external to systems.
  • filinks — a summary of links to FIs.
  • islinks — a summary of inter-switch links.
  • extislinks — a summary of inter-switch links external to systems.
  • slowlinks — a summary of links running slower than expected.
  • slowconfiglinks — a summary of links configured to run slower than supported, includes slowlinks.
  • slowconnlinks — a summary of links connected with mismatched speed potential, includes slowconfiglinks.
  • misconfiglinks — a summary of links configured to run slower than supported.
  • misconnlinks — a summary of links connected with mismatched speed potential.
  • errors — a summary of links whose errors exceed counts in the configuration file.
  • otherports — a summary of ports not connected to the fabric.
  • linear — a summary of linear forwarding data base (FDB) for each switch.
  • mcast — a summary of multicast FDB for each switch in the fabric.
  • mcgroups — a summary of multicast groups.

when you use it with -d, here's what you can get:

  • -d0 — shows the number of multicast groups
  • -d1 — shows a list of multicast groups
  • -d2 — shows a list of members per multicast group

this report can be used with option -X.

  • portusage — a summary of ports referenced in linear FDB for each switch, broken down by NodeType of DLID.
  • pathusage — a summary of number of FI to FI paths routed through each switch port.
  • treepathusage — analysis of number of FI to FI paths routed through each switch port for a FAT tree.
  • portgroups — a summary of adaptive routing port groups for each switch.
  • quarantinednodes — a summary of quarantined nodes.
  • validateroutes — validates all routes in the fabric.
  • validatevlroutes — validates all routes in the fabric using SLSC, SCSC, and SCVL tables.
  • validatepgs — validates all port groups in the fabric.
  • validatecreditloops — validates topology configuration of the fabric to identify any existing credit loops.
  • validatevlcreditloops — validates topology configuration of the fabric including SLSC, SCSC, and SCVL tables to identify any existing credit loops.
  • validatemcroutes — validates multicast routes of the fabric to identify loops in multicast forwarding tables and detect MFT-multicast membership inconsistencies.
  • vfinfo — a summary of virtual fabric (vFabric) information.
  • vfmember — a summary of vFabric membership information.
  • dgmember — a summary of DeviceGroup membership information.
  • verifyfis — compares fabric (or snapshot) FIs to supplied topology and identifies differences and omissions.
  • verifysws — compares fabric (or snapshot) switches to supplied topology and identifies differences and omissions.
  • verifynodes — returns verifyfis and verifysws reports.
  • verifysms — compares fabric (or snapshot) SMs to supplied topology and identifies differences and omissions.
  • verifylinks — compares fabric (or snapshot) links to supplied topology and identifies differences and omissions.
  • verifyextlinks — compares fabric (or snapshot) links to supplied topology and identifies differences and omissions. limits analysis to links external to systems.
  • verifyfilinks — compares fabric (or snapshot) links to supplied topology and identify differences and omissions. limits analysis to links to FIs.
  • verifyislinks — compares fabric (or snapshot) links to supplied topology and identify differences and omissions. limits analysis to inter-switch links.
  • verifyextislinks — compares fabric (or snapshot) links to supplied topology and identify differences and omissions. limits analysis to inter-switch links external to systems.
  • verifyall — returns verifyfis, verifysws, verifysms, and verifylinks reports.
  • all — returns comps, nodes, ious, links, extlinks, slowconnlinks, and errors reports.
  • route — traces route between -S and -D points.
  • bfrctrl — reports Buffer Control Tables for all ports.
  • snapshot — outputs snapshot of the fabric state for later use as snapshot_input. this implies -x. may not be combined with other reports. when selected, -F, -P, -H, and -N options are ignored.
  • topology — outputs the topology of the fabric for later use as topology_input. this implies -x. may not be combined with other reports. use with detail level 3 or more to get Port element under Node in output xml.
  • none — no report, useful to clear statistics.

🦋💅 Point Syntax 💅🦋

okay so when you're pointing at stuff, you use these syntaxes. each one is a little key to a specific thing. let's go:

  • gid:value — value is numeric port GID of form: subnet:guid.
  • lid:value — value is numeric LID.
  • lid:value:node — value is numeric LID, selects entire node with given LID.
  • lid:value:port:value2 — value is numeric LID of node, value2 is port number.
  • portguid:value — value is numeric port GUID.
  • nodeguid:value — value is numeric node GUID.
  • nodeguid:value1:port:value2 — value1 is numeric node GUID, value2 is port number.
  • iocguid:value — value is numeric IOC GUID.
  • iocguid:value1:port:value2 — value1 is numeric IOC GUID, value2 is port number.
  • systemguid:value — value is numeric system image GUID.
  • systemguid:value1:port:value2 — value1 is the numeric system image GUID, value2 is port number.
  • ioc:value — value is IOC Profile ID String (IOC Name).
  • ioc:value1:port:value2 — value1 is IOC Profile ID String (IOC Name), value2 is port number.
  • iocpat:value — value is glob pattern for IOC Profile ID String (IOC Name).
  • iocpat:value1:port:value2 — value1 is glob pattern for IOC Profile ID String (IOC Name), value2 is port number.
  • ioctype:value — value is IOC type (SRP or OTHER).
  • ioctype:value1:port:value2 — value1 is IOC type (SRP or OTHER); value2 is port number.
  • node:value — value is node description (node name).
  • node:value1:port:value2 — value1 is node description (node name), value2 is port number.
  • nodepat:value — value is glob pattern for node description (node name).
  • nodepat:value1:port:value2 — value1 is the glob pattern for the node description (node name), value2 is port number.
  • nodedetpat:value — value is glob pattern for node details.
  • nodedetpat:value1:port:value2 — value1 is the glob pattern for the node details, value2 is port number.
  • nodetype:value — value is node type (SW or FI).
  • nodetype:value1:port:value2 — value1 is node type (SW or FI), value2 is port number.
  • rate:value — value is string for rate (25g, 50g, 75g, 100g), omits switch mgmt port 0.
  • portstate:value — value is a string for state (down, init, armed, active, notactive, initarmed).
  • portphysstate:value — value is a string for PHYs state (polling, disabled, training, linkup, recovery, offline, test)
  • mtucap:value — value is MTU size (2048, 4096, 8192, 10240), omits switch mgmt port 0.
  • labelpat:value — value is glob pattern for cable label.
  • lengthpat:value — value is glob pattern for cable length.
  • cabledetpat:value — value is glob pattern for cable details.
  • cabinflenpat:value — value is glob pattern for cable info length.
  • cabinfvendnamepat:value — value is glob pattern for cable info vendor name.
  • cabinfvendpnpat:value — value is glob pattern for cable info vendor part number.
  • cabinfvendrevpat:value — value is glob pattern for cable info vendor revision.
  • cabinfvendsnpat:value — value is glob pattern for cable info vendor serial number.
  • cabinftype:value — value is either optical, passive_copper, active_copper, or unknown.
  • linkdetpat:value — value is glob pattern for link details.
  • portdetpat:value — value is glob pattern for port details.
  • sm — specifies the master subnet manager (SM).
  • smdetpat:value — value is glob pattern for SM details.
  • route:point1:point2 — specifies all ports along the routes between the two given points.
  • led:value — value is either on or off for LED port beacon.
  • linkqual:value — specifies the ports with a link quality equal to value.
  • linkqualLE:value — specifies the ports with a link quality less than or equal to value.
  • linkqualGE:value — specifies the ports with a link quality greater than or equal to value.
  • nodepatfile:FILENAME — specifies the name of file with the list of nodepats or node descriptions.
  • nodepairpatfile:FILENAME — specifies the name of file with the list of node pairs, separated by a colon.
  • ldr — specifies the ports with a non-zero link down reason or neighbor link down reason.
  • ldr:value — specifies the ports with a link down reason or neighbor link down reason equal to value.

✨ Examples ✨

okay bestie, so opareport can generate hundreds of different reports. The ones people actually run all the time? Say less. Here's the iconic cheat sheet, copy-paste responsibly:

opareport -o comps -d 3

opareport -o errors -o slowlinks

opareport -o nodes -F portguid:0x00117500a000447b

opareport -o nodes -F nodeguid:0x001175009800447b:port:1

opareport -o nodes -F nodeguid:0x001175009800447b

opareport -o nodes -F 'node:duster hfi1_0'

opareport -o nodes -F 'node:duster hfi1_0:port:1'

opareport -o nodes -F 'nodepat:d*'

opareport -o nodes -F 'nodepat:d*:port:1'

opareport -o nodes -F 'nodedetpat:compute*'

opareport -o nodes -F 'nodedetpat:compute*:port:1'

opareport -o nodes -F nodetype:FI

opareport -o nodes -F nodetype:FI:port:1

opareport -o nodes -F lid:1

opareport -o nodes -F led:on
opareport -o nodes -F led:off
opareport -o nodes -F lid:1:node

opareport -o nodes -F lid:1:port:2

opareport -o nodes -F gid:0xfe80000000000000:0x00117500a000447b

opareport -o nodes -F systemguid:0x001175009800447b

opareport -o nodes -F systemguid:0x001175009800447b:port:1

opareport -o nodes -F iocguid:0x00117501300001e0

opareport -o nodes -F iocguid:0x00117501300001e0:port:2

opareport -o nodes -F 'ioc:Chassis 0x001175005000010C, Slot 2, IOC 1'

opareport -o nodes -F 'ioc:Chassis 0x001175005000010C, Slot 2, IOC
       1:port:2'

opareport -o nodes -F 'iocpat:*Slot 2*'

opareport -o nodes -F 'iocpat:*Slot 2*:port:2'

opareport -o nodes -F ioctype:SRP
opareport -o nodes -F ioctype:SRP:port:2

opareport -o extlinks -F rate:100g

opareport -o extlinks -F portstate:armed

opareport -o extlinks -F portphysstate:linkup

opareport -o extlinks -F 'labelpat:S1345*'

opareport -o extlinks -F 'lengthpat:11m'

opareport -o extlinks -F 'cabledetpat:*hitachi*'

opareport -o extlinks -F 'linkdetpat:*core ISL*'

opareport -o extlinks -F 'portdetpat:*mgmt*'

opareport -o links -F mtucap:2048

opareport -o nodes -F sm

opareport -o nodes -F 'smdetpat:primary*'

opareport -o nodes -F 'route:node:duster hfi1_0:node:cuda hfi1_0'

opareport -o nodes -F 'route:node:duster hfi1_0:port:1:node:cuda
       hfi1_0:port:2'

opareport -o treepathusage -F nodepairpatfile:FILENAME

opareport -o pathusage -F nodepatfile:FILENAME

opareport -s -o snapshot > file

opareport -o topology > topology.xml

opareport -o errors -X file

opareport -s --begin "2 days ago"

opareport -s --begin "12:30" --end "14:00"

opareport -o linkinfo -x > file

💫 Other Information 💫

Okay so opareport also vibes with the Fabric Manager Performance Manager (PM)/Performance Manager Agent (PMA). When opareport detects the presence of a PM, it automatically issues any required PortCounter queries and clears to the PM to access the PMs running totals. If a PM is not detected, then opareport directly accesses the PMAs on all the nodes. The -M option can force access to the PMA even if a PM is present. periodt.

opareport takes advantage of these interfaces to obtain extensive information about the fabric from the subnet manager and the end nodes. Using this information, opareport is able to cross-reference it and produce analysis greatly beyond what any single subnet manager request could provide. As such, it exceeds the capabilities previously available in tools such as opasaquery and opafabricinfo. ate.

opareport obtains and displays counters from the Fabric Manager PM/PA or directly from the fabric PMAs using the -M option.

opareport internally cross-references all this information so its output can be in user-friendly form. Reports include GUIDs, LIDs, and names for components. Obviously, these reports are easiest to read if the end user has taken the time to provide unique names for all the components in the fabric (node names and IOC names). All Intel components support this capability. For hosts, the node names are automatically assigned based on the network host name of the server. For switches and line cards, the names can be assigned using the element managers for each component. iykyk.

Each run of opareport obtains up-to-date information from the fabric. At the start of the run opareport takes a few seconds to obtain all the fabric data, then it is output to stdout. The reports are sorted by GUIDs and other permanent information so they can be rerun in the future and produce output in the same order even if components have been rebooted. This is useful for comparison using simple tools like diff. opareport permits multiple reports to be requested for a single run (for example, one of each report type). slay.

By default, opareport uses the first active port on the local system. However, if the Management Node is connected to more than one fabric (for example, a subnet), the Intel(R) Omni-Path Host Fabric Interface (HFI) and port may be specified to select the fabric to analyze. main character energy.

💖 That's the whole vibe, bestie! This page is from Intel Corporation, copyright 2015-2019, and it's the opareport(8) manual. Go query that fabric and slay. ✨🦋

📦 where this page comes from