CSIT 24.06 - VPP Performance #
- TEST FRAMEWORK
- CSIT test environment version has been updated to ver. 15, see Environment Versioning.
- General Code Housekeeping: Ongoing code optimizations and bug fixes.
- VPP PERFORMANCE TESTS
- Added tests:
- Added memif+DMA tests; added 1518B and 4c memif testcases.
- Added nginx+DMA tests; added 2048B testcases.
- Added IPsec hwasync tests to 3n-icxd and 3n-snr.
- Added IPsec tests to cover more encryption algorithms and other settings.
- Added more SOAK tests.
- Added selected 6-port tests for 3na-spr.
- Edited tests:
- Selected single-flow tests now use single worker even if SMT is on.
- IPsecHW tests now use rxq ratio of 2.
- This means one worker reads only from one of two ports.
- This workaround avoids some inefficiencies,
- but still does not reach the expected performance on 3nb-spr.
- 1518B tests with encapsulation overhead now properly use no-multi-seg.
- Added TX checksum offload to hoststack tests missing it.
- Added tests:
- PRESENTATION AND ANALYTICS LAYER
- Detailed views added to comparison tables.
Known Issues #
These are issues that cause test failures or otherwise limit usefulness of CSIT testing.
New #
Any issue listed here may have been present also in a previous release, but was not detected/recognized/reported enough back then. Also, issues previously thought fixed but now reopened are listed here.
| # | JiraID | Issue Description |
|---|---|---|
| 1 | CSIT-1877 | 3n-tsh: VM tests too slow to boot VM, rarely, despite increased timeout. |
| 2 | CSIT-1884 | 2n-clx, 2n-icx: All NAT44DET NDR PDR IMIX over 1M sessions BIDIR tests failing to create enough sessions. |
| 3 | VPP-2118 | 3n spr: Unusable performance of ipsec tests with SHA_256_128. |
| 4 | CSIT-1960 | 2n-zn2: Geneve sometimes loses one direction of traffic. |
| 5 | CSIT-1961 | Some tests have too long ramp-up trials. |
| 6 | CSIT-1962 | 3n-icx hoststack: Udpquicscale tests sometimes fail with various symptoms. |
| 7 | CSIT-1963 | 3n-icxd: Various symptoms pointing to hardware (cable/nic/driver) issues. |
| 8 | CSIT-1964 | 3nb-spr: Wireguardhw tests are likely to crash. |
| 9 | CSIT-1965 | Occasional failure on 1518B CX5: Trex failed to send message. |
| 10 | VPP-2121 | sw_interface_add_del_address: avf process node failed to reply in 5 seconds |
| 11 | CSIT-1966 | 3n-snr: Increased heap size in ipsec policy tests prevents VPP from starting. |
| 12 | CSIT-1967 | 3na-spr: Unable to configure large MTU for 9000B tests. |
Previous #
Issues reported in previous releases which still affect the current results.
| # | JiraID | Issue Description |
|---|---|---|
| 1 | CSIT-1785 | NAT44ED tests failing to establish all TCP sessions. At least for max scale, in allotted time (limited by session 500s timeout) due to worse slow path performance than previously measured and calibrated for. CSIT removed the max scale NAT tests to avoid this issue. |
| 2 | CSIT-1795 | Ocassionally not all DET44 sessions have been established: 4128767 != 4128768. |
| 3 | CSIT-1802 | All testbeds: AF-XDP - NDR tests failing from time to time on small loss. |
| 4 | CSIT-1804 | 3n-tsh: NDR fails on ierrors. |
| 5 | CSIT-1845 | AVF 9000B any ndrpdr test may start failing due to packets not arriving in one or both directions. |
| 6 | CSIT-1886 | 3n: Wireguard tests with 100 and more tunnels are failing PDR criteria. |
| 7 | CSIT-1892 | 3n-alt: Unexpected two-band structure of ipsec and vxlan. |
| 8 | CSIT-1896 | Depending on topology, l3fwd avoids dut-dut link. |
| 9 | CSIT-1901 | 2n-icx 3n-icx: Trex may report negative ipackets on high-performance AVF trial. |
| 10 | CSIT-1904 | DPDK 23.03 testpmd startup fails on some testbeds. |
| 11 | CSIT-1906 | Zero traffic with cx7 rdma. Testing migrated to mlx5-core on all Mellanox NICs. |
| 12 | VPP-2077 | IP fragmentation: running_fragment_id is not thread safe. Causes reduced performance and failures in gtpu reassembly tests. |
| 13 | CSIT-1914 | TRex does not produce latency data on ICE NICs. |
| 14 | CSIT-1915 | The 2n-icx testbeds do not have the same performance. |
| 15 | CSIT-1916 | Poor CPU scaling on 2n-zn2 RDMA. |
| 16 | CSIT-1917 | TRex STL performance is unstable at high pps due to unsent packets. |
| 17 | CSIT-1922 | 2n-tx2: AF_XDP MRR failures. On other testbeds MRR regressions and PDR failures. |
| 18 | CSIT-1924 | An l3fwd error in 200Ge2P1Cx7Veat-Mlx5 test with 9000B. |
| 19 | CSIT-1929 | Lossy trials in nat udp mlx5 tests. |
| 20 | CSIT-1935 | rls2310: Zero traffic reported in udpquic tests due to session close errors. |
| 21 | CSIT-1936 | TRex occasionally sees link down in L2 perf tests. |
| 22 | CSIT-1938 | 3n-alt: High scale ipsec policy tests may crash VPP. |
| 23 | CSIT-1941 | TRex may wrongly detect link bandwidth. |
| 24 | CSIT-1942 | 3nb-spr hoststack: Interface not up after first test. |
| 25 | CSIT-1943 | IMIX 4c tests may fail PDR due to ~10% loss. |
| 26 | CSIT-1944 | Memif LXC: unrecognized option ‘–no-validate’. |
| 27 | VPP-2090 | MRR < PDR: DPDK plugin with MLX5 driver does not read full queue. |
| 28 | CSIT-1946 | Ipsec hwasync fails with large scale and multiple queues. |
| 29 | CSIT-1947 | Rare VPP crash in nat avf tests. |
| 30 | VPP-2091 | Memif crashes VPP in container with jumbo frames. |
| 31 | CSIT-1948 | NICs do not consistently distribute tunnels over RXQs depending on model or plugin. |
| 32 | CSIT-1950 | 9000B tests with encap overhead and non-dpdk plugins see fragmented packets. |
| 33 | CSIT-1951 | Combination of AVF and vhost drops all 9000B packets. |
| 34 | CSIT-1953 | 3n-icx 3nb-spr: Failed to enable GTPU offload RX. |
Fixed #
Issues reported in previous releases which were fixed in this release:
| # | JiraID | Issue Description |
|---|---|---|
| 1 | CSIT-1782 | Multicore AVF tests are failing when trying to create interface. Frequency is reduced by CSIT workaround, but occasional failures do still happen. |
| 2 | CSIT-1881 | 2n-icx: NFV density tests ocassionally breaks VPP which fails to start. |
| 3 | CSIT-1921 | Two-band structure in SRv6, causes PDR failure in rare cases. |
| 4 | CSIT-1939 | 3na-spr, 2n-zn2: VPP fails to start in first test cases. |
Root Cause Analysis for Regressions #
List of RCAs in CSIT 24.06 for VPP performance regressions. Not listing differences caused by known issues (uneven worker load due to randomized RSS or other per-worker issues). Also not listing tests which historically show large performance variance.
Contrary to issues, these genuine regressions do not limit usefulness of CSIT testing. So even if they are not fixed (e.g. when the regression is an expected consequence of added functionality), they will not be re-listed in the next release report.
| # | JiraID | Issue Description |
|---|---|---|
| 1 | CSIT-1958 | 3n-icx 3na-spr: Mixed performance impact of 40373 around 2024-04-20. |
| 2 | CSIT-1959 | Explain change in c6in performance. |