From 01850bee3d854b46052d8ba4a7faba3b63f9e89e Mon Sep 17 00:00:00 2001 From: Maciek Konstantynowicz Date: Mon, 13 Jul 2020 18:14:16 +0200 Subject: report: 2n-clx, 2n-skx, 3n-skx RCA edits in comparison tables Change-Id: I9810b0b33355d880b606d7402564a3150d12397c Signed-off-by: Maciek Konstantynowicz (cherry picked from commit 7d81bdcc650170eaa304defbf6b3b006245697fb) --- .../presentation/rca/rca-2n-clx-2t1c-pdr-rca1.yaml | 49 +++++++++--------- .../presentation/rca/rca-2n-skx-2t1c-ndr-rca1.yaml | 49 +++++++++--------- .../presentation/rca/rca-2n-skx-2t1c-pdr-rca1.yaml | 59 ++++++++++------------ .../presentation/rca/rca-3n-skx-2t1c-pdr-rca1.yaml | 37 +++++++------- 4 files changed, 92 insertions(+), 102 deletions(-) diff --git a/resources/tools/presentation/rca/rca-2n-clx-2t1c-pdr-rca1.yaml b/resources/tools/presentation/rca/rca-2n-clx-2t1c-pdr-rca1.yaml index 21125d18ba..136c23f88d 100644 --- a/resources/tools/presentation/rca/rca-2n-clx-2t1c-pdr-rca1.yaml +++ b/resources/tools/presentation/rca/rca-2n-clx-2t1c-pdr-rca1.yaml @@ -22,45 +22,44 @@ # see https://yaml-multiline.info/ footnote: " - [1] DONE Regression in ACL tests. The cause is the CSIT part of - an API change, Gerrit 26112. Instead of failing, the misconfigured VPP - passes the test, albeit with reduced performance. [CSIT-1739] - [2] DONE Regression in tests using random flows. - The cause is Gerrit 26456, unpinning random seed, needed to make TRex workers - generate different sequences of packets. [CSIT-1731] + [1] DONE Regression in ACL tests due to Gerrit 26112. With CSIT + using changed VPP API, instead of failing, misconfigured VPP passes + the test, albeit with reduced performance. [CSIT-1739] - [3] DONE Regression in base dot1q-l2xcbase suites. The cause is Gerrit 25363, - which unified dot1q handling across various tests. - Previously, this one suite was different (only handling dot1q on one port). + [2] DONE Regression in tests using random flow sequencing due to + Gerrit 26456. Unpinned random seed to make TRex workers generate + different sequences of packets. [CSIT-1731] - [5] DONE Progression in performance of tests running VPP (l2xc) inside VM. - Consequence of a fix making worker placement deterministic, Gerrit 25309. + [3] DONE Regression in base dot1q-l2xcbase suites due to Gerrit + 25363. Unified dot1q handling across various tests. Previously, + this one suite was different, using dot1q on one port only. + + [5] DONE Progression in tests with VPP (l2xc) inside a VM due to + Gerrit 25309. Make VPP in VM worker placement deterministic. [CSIT-1738] [7] POSTPONED Possible anomaly, but stdev of results is high. - Currently, the performance differences are considered to probably be just noise. + Performance differences considered to probably be just noise. - [9] OPEN Progression in ip6 scale tests. Multiple candidate changes may have - affect code efficiency, but they have also caused the tests to fail - for a period of time. [CSIT-1735] + [9] OPEN Progression in all ip6 scale tests. Changes to vppinfra + bihash may have improved code efficiency. [CSIT-1735] [10] OPEN Progression in ip6 base tests with dot1q. [CSIT-1746] - [11] OPEN Progression in memif tests, mainly in avf-eth-l2xcbase-eth-2memif-1dcr. - Not sure whether all memif tests are affected, nor what the cause is. - [CSIT-1744] + [11] OPEN Progression in memif tests, mainly in + avf-eth-l2xcbase-eth-2memif-1dcr. [CSIT-1744] [12] OPEN Progression in some l2bdscale tests. [CSIT-1745] - [13] DONE Stateless ACL test have the same performance in 2005.0-4 as in - 2001.0-2. The difference shown here is inflated by an API change, see [1]. - [CSIT-1739] + [13] DONE Stateless ACL test have the same performance in 2005.0-4 + as in 2001.0-2. The difference shown here is inflated by an API + change, see [1]. [CSIT-1739] - [14] DONE Stateful ACL test have better performance in 2005.0-4 as in - 2001.0-2. The VPP part of API change, Gerrit 26184, made VPP faster. - The difference shown here is further inflated by the API change, see [1]. - [CSIT-1739] + [14] DONE Stateful ACL test have better performance in 2005.0-4 as + in 2001.0-2. The VPP part of API change, Gerrit 26184, made VPP + faster. The difference shown here is further inflated by the API + change, see [1]. [CSIT-1739] [15] OPEN Progression in nat44 test (not scale ones). [CSIT-1747] diff --git a/resources/tools/presentation/rca/rca-2n-skx-2t1c-ndr-rca1.yaml b/resources/tools/presentation/rca/rca-2n-skx-2t1c-ndr-rca1.yaml index 8d323a48af..2274748d78 100644 --- a/resources/tools/presentation/rca/rca-2n-skx-2t1c-ndr-rca1.yaml +++ b/resources/tools/presentation/rca/rca-2n-skx-2t1c-ndr-rca1.yaml @@ -35,44 +35,41 @@ # see https://yaml-multiline.info/ footnote: " - [1] DONE Regression in ACL tests. The cause is the CSIT part of - an API change, Gerrit 26112. Instead of failing, the misconfigured VPP - passes the test, albeit with reduced performance. [CSIT-1739] + [1] DONE Regression in ACL tests due to Gerrit 26112. With CSIT + using changed VPP API, instead of failing, misconfigured VPP passes + the test, albeit with reduced performance. [CSIT-1739] - [2] DONE Regression in tests using random flows. - The cause is Gerrit 26456, unpinning random seed, needed to make TRex workers - generate different sequences of packets. [CSIT-1731] + [2] DONE Regression in tests using random flow sequencing due to + Gerrit 26456. Unpinned random seed to make TRex workers generate + different sequences of packets. [CSIT-1731] - [5] DONE Progression in performance of tests running VPP (l2xc) inside VM. - Consequence of a fix making worker placement deterministic, Gerrit 25309. + [5] DONE Progression in tests with VPP (l2xc) inside a VM due to + Gerrit 25309. Make VPP in VM worker placement deterministic. [CSIT-1738] [7] POSTPONED Possible anomaly, but stdev of results is high. - Currently, the performance differences are considered to probably be just noise. + Performance differences considered to probably be just noise. - [8] DONE Regressions in some tests with VPP in VM. - Tests using l2xcbase as vswitch have small enough stdev for bisect. - The switch to clang-9 compiler, Gerrit 26744, has been identified as the cause. - [CSIT-1741] + [8] DONE Regressions in some tests with VPP in VM due to Gerrit + 26744. Switch to clang-9 compiler identified as cause. Tests using + l2xcbase as vswitch with small stdev used for bisect. [CSIT-1741] - [9] OPEN Progression in ip6 scale tests. Multiple candidate changes may have - affect code efficiency, but they have also caused the tests to fail - for a period of time. [CSIT-1735] + [9] OPEN Progression in all ip6 scale tests. Changes to vppinfra + bihash may have improved code efficiency. [CSIT-1735] - [11] OPEN Progression in memif tests, mainly in avf-eth-l2xcbase-eth-2memif-1dcr. - Not sure whether all memif tests are affected, nor what the cause is. - [CSIT-1744] + [11] OPEN Progression in memif tests, mainly in + avf-eth-l2xcbase-eth-2memif-1dcr. [CSIT-1744] [12] OPEN Progression in some l2bdscale tests. [CSIT-1745] - [13] DONE Stateless ACL test have the same performance in 2005.0-4 as in - 2001.0-2. The difference shown here is inflated by an API change, see [1]. - [CSIT-1739] + [13] DONE Stateless ACL test have the same performance in 2005.0-4 + as in 2001.0-2. The difference shown here is inflated by an API + change, see [1]. [CSIT-1739] - [14] DONE Stateful ACL test have better performance in 2005.0-4 as in - 2001.0-2. The VPP part of API change, Gerrit 26184, made VPP faster. - The difference shown here is further inflated by the API change, see [1]. - [CSIT-1739] + [14] DONE Stateful ACL test have better performance in 2005.0-4 as + in 2001.0-2. The VPP part of API change, Gerrit 26184, made VPP + faster. The difference shown here is further inflated by the API + change, see [1]. [CSIT-1739] [15] OPEN Progression in nat44 test (not scale ones). [CSIT-1747] diff --git a/resources/tools/presentation/rca/rca-2n-skx-2t1c-pdr-rca1.yaml b/resources/tools/presentation/rca/rca-2n-skx-2t1c-pdr-rca1.yaml index 496221dd98..8cc7ee352d 100644 --- a/resources/tools/presentation/rca/rca-2n-skx-2t1c-pdr-rca1.yaml +++ b/resources/tools/presentation/rca/rca-2n-skx-2t1c-pdr-rca1.yaml @@ -23,51 +23,48 @@ # see https://yaml-multiline.info/ footnote: " - [1] DONE Regression in ACL tests. The cause is the CSIT part of - an API change, Gerrit 26112. Instead of failing, the misconfigured VPP - passes the test, albeit with reduced performance. [CSIT-1739] + [1] DONE Regression in ACL tests due to Gerrit 26112. With CSIT + using changed VPP API, instead of failing, misconfigured VPP passes + the test, albeit with reduced performance. [CSIT-1739] - [2] DONE Regression in tests using random flows. - The cause is Gerrit 26456, unpinning random seed, needed to make TRex workers - generate different sequences of packets. [CSIT-1731] + [2] DONE Regression in tests using random flow sequencing due to + Gerrit 26456. Unpinned random seed to make TRex workers generate + different sequences of packets. [CSIT-1731] - [3] DONE Regression in base dot1q-l2xcbase suites. The cause is Gerrit 25363, - which unified dot1q handling across various tests. - Previously, this one suite was different (only handling dot1q on one port). + [3] DONE Regression in base dot1q-l2xcbase suites due to Gerrit + 25363. Unified dot1q handling across various tests. Previously, + this one suite was different, using dot1q on one port only. - [4] DONE Progression in high performance tests, caused by - TRex setting improvement, Gerrit 26445. [CSIT-1740] + [4] DONE Progression in high performance tests due to Gerrit 26445. + TRex setting improvement. [CSIT-1740] - [5] DONE Progression in performance of tests running VPP (l2xc) inside VM. - Consequence of a fix making worker placement deterministic, Gerrit 25309. + [5] DONE Progression in tests with VPP (l2xc) inside a VM due to + Gerrit 25309. Make VPP in VM worker placement deterministic. [CSIT-1738] [7] POSTPONED Possible anomaly, but stdev of results is high. - Currently, the performance differences are considered to probably be just noise. + Performance differences considered to probably be just noise. - [8] DONE Regressions in some tests with VPP in VM. - Tests using l2xcbase as vswitch have small enough stdev for bisect. - The switch to clang-9 compiler, Gerrit 26744, has been identified as the cause. - [CSIT-1741] + [8] DONE Regressions in some tests with VPP in VM due to Gerrit + 26744. Switch to clang-9 compiler identified as cause. Tests using + l2xcbase as vswitch with small stdev used for bisect. [CSIT-1741] - [9] OPEN Progression in ip6 scale tests. Multiple candidate changes may have - affect code efficiency, but they have also caused the tests to fail - for a period of time. [CSIT-1735] + [9] OPEN Progression in all ip6 scale tests. Changes to vppinfra + bihash may have improved code efficiency. [CSIT-1735] - [11] OPEN Progression in memif tests, mainly in avf-eth-l2xcbase-eth-2memif-1dcr. - Not sure whether all memif tests are affected, nor what the cause is. - [CSIT-1744] + [11] OPEN Progression in memif tests, mainly in + avf-eth-l2xcbase-eth-2memif-1dcr. [CSIT-1744] [12] OPEN Progression in some l2bdscale tests. [CSIT-1745] - [13] DONE Stateless ACL test have the same performance in 2005.0-4 as in - 2001.0-2. The difference shown here is inflated by an API change, see [1]. - [CSIT-1739] + [13] DONE Stateless ACL test have the same performance in 2005.0-4 + as in 2001.0-2. The difference shown here is inflated by an API + change, see [1]. [CSIT-1739] - [14] DONE Stateful ACL test have better performance in 2005.0-4 as in - 2001.0-2. The VPP part of API change, Gerrit 26184, made VPP faster. - The difference shown here is further inflated by the API change, see [1]. - [CSIT-1739] + [14] DONE Stateful ACL test have better performance in 2005.0-4 as + in 2001.0-2. The VPP part of API change, Gerrit 26184, made VPP + faster. The difference shown here is further inflated by the API + change, see [1]. [CSIT-1739] [15] OPEN Progression in nat44 test (not scale ones). [CSIT-1747] diff --git a/resources/tools/presentation/rca/rca-3n-skx-2t1c-pdr-rca1.yaml b/resources/tools/presentation/rca/rca-3n-skx-2t1c-pdr-rca1.yaml index ab578f18a1..a132a0a232 100644 --- a/resources/tools/presentation/rca/rca-3n-skx-2t1c-pdr-rca1.yaml +++ b/resources/tools/presentation/rca/rca-3n-skx-2t1c-pdr-rca1.yaml @@ -21,39 +21,36 @@ imix-2t1c-ethip4ipsec4tnlsw-ip4base-int-aes256gcm: 6 # see https://yaml-multiline.info/ footnote: " - [4] DONE Progression in high performance tests, caused by - TRex setting improvement, Gerrit 26445. [CSIT-1740] + [4] DONE Progression in high performance tests due to Gerrit 26445. + TRex setting improvement. [CSIT-1740] - [5] DONE Progression in performance of tests running VPP (l2xc) inside VM. - Consequence of a fix making worker placement deterministic, Gerrit 25309. + [5] DONE Progression in tests with VPP (l2xc) inside a VM due to + Gerrit 25309. Make VPP in VM worker placement deterministic. [CSIT-1738] - [6] OPEN Regression in CSIT, one specific test showing too small stdev - to be considered noise. Tentatively marking also other ipsec anomalies - with this, although their stdev are high enough to be explained - as maybe just noise. [CSIT-1742] + [6] OPEN Regression in CSIT, one specific test showing too small + stdev to be considered noise. Tentatively marking also other ipsec anomalies + with this, albeit they show higher stdev. [CSIT-1742] [7] POSTPONED Possible anomaly, but stdev of results is high. - Currently, the performance differences are considered to probably be just noise. + Performance differences considered to probably be just noise. - [8] DONE Regressions in some tests with VPP in VM. - Tests using l2xcbase as vswitch have small enough stdev for bisect. - The switch to clang-9 compiler, Gerrit 26744, has been identified as the cause. - [CSIT-1741] + [8] DONE Regressions in some tests with VPP in VM due to Gerrit + 26744. Switch to clang-9 compiler identified as cause. Tests using + l2xcbase as vswitch with small stdev used for bisect. [CSIT-1741] - [9] OPEN Progression in ip6 scale tests. Multiple candidate changes may have - affect code efficiency, but they have also caused the tests to fail - for a period of time. [CSIT-1735] + [9] OPEN Progression in all ip6 scale tests. Changes to vppinfra + bihash may have improved code efficiency. [CSIT-1735] [10] OPEN Progression in ip6 base tests with dot1q. [CSIT-1746] - [11] OPEN Progression in memif tests, mainly in avf-eth-l2xcbase-eth-2memif-1dcr. - Not sure whether all memif tests are affected, nor what the cause is. - [CSIT-1744] + [11] OPEN Progression in memif tests, mainly in + avf-eth-l2xcbase-eth-2memif-1dcr. [CSIT-1744] [12] OPEN Progression in some l2bdscale tests. [CSIT-1745] - [16] OPEN Progression in IPsec tests, only for aes128cbc-hmac512sha. [CSIT-1748] + [16] OPEN Progression in IPsec tests, only for aes128cbc-hmac512sha. + [CSIT-1748] [17] OPEN Progression in some vhost tests (dpdk in VM). [CSIT-1749] " -- cgit 1.2.3-korg