ceph/src/spdk/test/nvmf/test_plan.md

   1 # SPDK nvmf_tgt test plan
   2
   3 ## Objective
   4 The purpose of these tests is to verify correct behavior of SPDK NVMe-oF
   5 feature.
   6 These tests are run either per-commit or as nightly tests.
   7
   8 ## Configuration
   9 All tests share the same basic configuration file for SPDK nvmf_tgt to run.
  10 Static configuration from config file consists of setting number of per session
  11 queues and enabling RPC for further configuration via RPC calls.
  12 RPC calls used for dynamic configuration consist:
  13 - creating Malloc backend devices
  14 - creating Null Block backend devices
  15 - constructing NVMe-oF subsystems
  16 - deleting NVMe-oF subsystems
  17
  18 ### Tests
  19
  20 #### Test 1: NVMe-oF namespace on a Logical Volumes device
  21 This test configures a SPDK NVMe-oF subsystem backed by logical volume
  22 devices and uses FIO to generate I/Os that target those subsystems.
  23 The logical volume bdevs are backed by malloc bdevs.
  24 Test steps:
  25 - Step 1: Assign IP addresses to RDMA NICs.
  26 - Step 2: Start SPDK nvmf_tgt application.
  27 - Step 3: Create malloc bdevs.
  28 - Step 4: Create logical volume stores on malloc bdevs.
  29 - Step 5: Create 10 logical volume bdevs on each logical volume store.
  30 - Step 6: Create NVMe-oF subsystems with logical volume bdev namespaces.
  31 - Step 7: Connect to NVMe-oF susbsystems with kernel initiator.
  32 - Step 8: Run FIO with workload parameters: blocksize=256k, iodepth=64,
  33 workload=randwrite; varify flag is enabled so that FIO reads and verifies
  34 the data written to the logical device. The run time is 10 seconds for a
  35 quick test an 10 minutes for longer nightly test.
  36 - Step 9: Disconnect kernel initiator from NVMe-oF subsystems.
  37 - Step 10: Delete NVMe-oF subsystems from configuration.
  38
  39 ### Compatibility testing
  40
  41 - Verify functionality of SPDK `nvmf_tgt` with Linux kernel NVMe-oF host
  42   - Exercise various kernel NVMe host parameters
  43     - `nr_io_queues`
  44     - `queue_size`
  45   - Test discovery subsystem with `nvme` CLI tool
  46     - Verify that discovery service works correctly with `nvme discover`
  47     - Verify that large responses work (many subsystems)
  48
  49 ### Specification compliance
  50
  51 - NVMe base spec compliance
  52   - Verify all mandatory admin commands are implemented
  53     - Get Log Page
  54     - Identify (including all mandatory CNS values)
  55       - Identify Namespace
  56       - Identify Controller
  57       - Active Namespace List
  58       - Allocated Namespace List
  59       - Identify Allocated Namespace
  60       - Attached Controller List
  61       - Controller List
  62     - Abort
  63     - Set Features
  64     - Get Features
  65     - Asynchronous Event Request
  66     - Keep Alive
  67   - Verify all mandatory NVM command set I/O commands are implemented
  68     - Flush
  69     - Write
  70     - Read
  71   - Verify all mandatory log pages
  72     - Error Information
  73     - SMART / Health Information
  74     - Firmware Slot Information
  75   - Verify all mandatory Get/Set Features
  76     - Arbitration
  77     - Power Management
  78     - Temperature Threshold
  79     - Error Recovery
  80     - Number of Queues
  81     - Write Atomicity Normal
  82     - Asynchronous Event Configuration
  83   - Verify all implemented commands behave as required by the specification
  84 - Fabric command processing
  85   - Verify that Connect commands with invalid parameters are failed with correct response
  86     - Invalid RECFMT
  87     - Invalid SQSIZE
  88     - Invalid SUBNQN, HOSTNQN (too long, incorrect format, not null terminated)
  89     - QID != 0 before admin queue created
  90     - CNTLID != 0xFFFF (static controller mode)
  91   - Verify that non-Fabric commands are only allowed in the correct states
  92
  93 ### Configuration and RPC
  94
  95 - Verify that invalid NQNs cannot be configured via conf file or RPC