@dotnet/eng-first are there currently any known issues with the CI on 2.0 branch? It doesn't seem to have started.

@krwq The dev/release/2.0.0 branch isn't a tracked branch. Is this where you meant to make your PR?

Currently I do not understand why exactly RedHat 6.9 uses code path which is conditioned for Win10/OSX15+/Ubuntu14.04+/Fedora/Debian>8

on Linux, it depends on what version of ICU is used on such distro with a specific version. On Windows, Windows is changing/adding globalization data almost every release.
So Looks Red Hat 6.9 is using ICU version which is used in Win10/OSX15+/Ubuntu14.04+/Fedora/Debian>8

👍

@tarekgh seems like I have swapped Expected value with Actual when reading and reversed condition (should be || Rhel6.9)

seems like I have swapped Expected value with Actual when reading and reversed condition (should be || Rhel6.9)

The expected value for the test case is [3, 2]. so I think using !redhat6.9 is correct.

@tarekgh I believe [3] is correct expected value - that is what Win10/ICU 55+ is using (also what you wrote above) - !RedHat6.9 wouldn't change test result as we currently already expect [3,2] and seems like RHEL6.9 used by CI have ICU 55+. (note that UrINNumberGroupSizes returns expected value not actual and there is no reason any of the conditions in the if statement would be met)
https://mc.dot.net/#/product/netcore/200/source/official~2Fcorefx~2Frelease~2F2.0.0~2F/type/test~2Ffunctional~2Fcli~2F/build/20170927.01

Closing this in favor of #24340 - this PR is improperly targeting dev/release/2.0.0

@krwq
the expected value is Expected: Int32[] [3, 2]

2017-09-27 13:43:11,451: INFO: proc(54): run_and_log_output: Output:    System.Globalization.Tests.NumberFormatInfoNumberGroupSizes.NumberGroupSizes_Get(format: NumberFormatInfo { CurrencyDecimalDigits = 2, CurrencyDecimalSeparator = \".\", CurrencyGroupSeparator = \",\", CurrencyGroupSizes = [3, 2], CurrencyNegativePattern = 9, ... }, expected: [3, 2]) [FAIL]
2017-09-27 13:43:11,453: INFO: proc(54): run_and_log_output: Output:       Assert.Equal() Failure
2017-09-27 13:43:11,453: INFO: proc(54): run_and_log_output: Output:       Expected: Int32[] [3, 2]
2017-09-27 13:43:11,453: INFO: proc(54): run_and_log_output: Output:       Actual:   Int32[] [3]
2017-09-27 13:43:12,240: INFO: proc(54): run_and_log_output: Output: Finished:    System.Globalization.Tests

@tarekgh true, I think we got both confused by this condition:
UrINNumberGroupSizes - returns expected value [3] when Win10 or ICU 55+ otherwise [3,2]
RHEL 6.9 - CI uses ICU 55+ and the actual value in the failing test is [3] as it should be
current version of the code running on RHEL 6.9 is expecting [3,2] because none of the conditions is satisfied for RHEL and we get into the else part ([3,2]).