Changes in CSIT-1810¶
- VPP PERFORMANCE TESTS
- BMRR Throughput: MRR (Maximum Receive Rate) test code has been updated with a configurable burst MRR parameters: trial duration and number of trials in a single burst. Enables a new Burst MRR (BMRR for short) methodology for more precise performance trending. See updated performance_test_methodology section for more details.
- 2n-skx tests: Added performamce tests for 2n-skx (2-Node Xeon Skylake) testbeds: focus on baseline and scale tests, including VM vhost and Container memif tests.
- 3n-skx tests: Added performamce tests for 3n-skx (3-Node Xeon Skylake) testbeds: VM vhost and Container memif tests.
- VXLAN Scale Tests: Added performamce tests for VXLAN scale with dot1q and VPP L2BD.
- AVF Driver Tests: Added performamce tests for i40e AVF driver on VPP, no DPDK required.
- QAT: Fixed reoccuring issues with QAT crypto accelerator cards.
- VM Vhost Virtio Params Combinations: Added performance tests for VM vhost with different virtio parameters combinations: indirect buffers, mergeable buffers.
- K8s/Ligato in Trending: Added K8s/Ligato Container memif tests to daily trending.
- TEST FRAMEWORK
- Experimental Soak Tests: Added performamce soak tests framework code for extended time duration tests and throughput discovery at given PLR and at give total test time e.g. minutes, hours, days, weeks, months, years. See updated performance_test_methodology section for more details.
- Trending Tests BMRR: Used new Burst MRR (BMRR) tests for daily trending.
- Per VPP Patch Performance Checks: Per VPP gerrit patch vs. parent performance tests, anomaly detection and no verify voting (-1/0/+1) yet. Manual trigger only. Not “marketed” to FD.io community yet to avoid excessive LFN FD.io physical performance testbed blocking.
- Patch-on-Patch Infra: Added capability to run performance tests using CSIT gerrit patch code testing VPP gerrit patch code, i.e. before any VPP and/or CSIT code is merged into git branch.
- CSIT PAPI Support: Initial implementation of PAPI L1 KWs in CSIT using VPP Python bindings. Required for migraing away from VAT. Very few L1 KWs implemented (“show version”, “show interfaces”).
- General Code Housekeeping: Ongoing RF keywords optimizations, removal of redundant RF keywords.
- PRESENTATION AND ANALYTICS LAYER
- Graphs Layout Improvements: Improved performance graphs layout for better readibility and maintenance: test grouping, axis labels, descriptions, other informative decoration. Master report generated. 744 graphs(!)
- Performance Trending: Further improvements of continuous performance trending, anomaly detection and analysis.
- 3n-dnv Tests (3rd Party): Published performance tests for 3n- dnv (3-Node Atom Denverton) from 3rd party testbeds running FD.io CSIT-1810 automated testing code. Only graphs for Packet Throughput and Speedup Multi-core and not for Packet Latency were published as there are no results for Packet Latency available.
List of known issues in CSIT-1810 for VPP performance tests:
|1||CSIT-570||Sporadic (1 in 200) NDR discovery test failures on x520. DPDK reporting rx-errors, indicating L1 issue. Suspected issue with HW combination of X710-X520 in LF testbeds. Not observed outside of LF testbeds.|
|2||CSIT-1234||VPP IPSecHW scale interface mode 1core, low NDR and PDR 64B throughput in 3n-hsw testbeds, in CSIT-18.07 vs. CSIT-18.04. ip4ipsecscale1000tnl-ip4base-int 1core CSIT-18.07/18.04 relative change: NDR -31%, PDR -32%, MRR -38%.|
|3||CSIT-1242||VPP xl710 ip4base test 1core, low NDR and PDR 64B throughput in 3n-hsw testbeds, in CSIT-18.07 vs. CSIT-18.04. xl710 ip4base 1core CSIT-18.07/18.04 relative change: NDR -29%, high stdev.|
|4||CSIT-1243||VPP nat44 base test 2core, low NDR and PDR 64B throughput in 3n-skx testbeds, compared to 3n-hsw testbeds. ip4base-nat44 2core 3n-skx/3n-hsw relative change: NDR -19%, PDR -22%.|
|5||CSIT-1246||Ligato K8S orchestrated tests are failing due to incompatibility of the latest released Ligato vpp-agent with VPP-18.07. Past vpp-agent releases are not compatible either.|