After the landing of LU-13133 both test_21a and test_21b are still failing regularly, but with a different error message. This is still happening on the latest master - 58/60 failures in the past 3 days. It looks like both review-dne-selinux and review-dne-selinux-ssk are both failing a majority of the times, though there is about one or two passes per day for some reason.
It looks like these are the only two subtest failures remaining for sanity-selinux in the past few days, so if they are excluded or fixed then we might finally be able to start enforcing this test session so that it stops getting new regressions.
Andreas Dilger (adilger@whamcloud.com) uploaded a new patch: https://review.whamcloud.com/37899
Subject: LU-13156 tests: skip sanity-selinux test_21a+21b
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 8351af09c42d6959677c2d15351c50859a615c88
Patch https://review.whamcloud.com/37224 was landed to fix sanity-selinux test_21a and test_21b, so they should not be skipped.
The 37224 patch was landed on 2020-02-24, but these subtests are still failing regularly so there appears to be a different problem (maybe exposed once the configuration issue was resulted?) For example, patch https://review.whamcloud.com/37620 commit 1071a24b57a34dc87e704ac9cc99132700d0d0a5 is based on the current tip of master but still failed both of these tess.
https://testing.whamcloud.com/test_sessions/461de6f9-30fe-4357-8088-489cd8295f29
Oops my bad. I totally misread this ticket 
You are right, this issue with sanity-selinux test_21a and test_21b is different from the one fixed by patch https://review.whamcloud.com/37224. Before patch https://review.whamcloud.com/37224, the client connection was working without sepol, which was not ok. Now that this patch has landed, we realize that sepol never gets updated, which is the source of the problem.
I think it is ok to skip those 2 tests for now, as sepol is not widely used.
Oleg Drokin (green@whamcloud.com) merged in patch https://review.whamcloud.com/37899/
Subject: LU-13156 tests: skip sanity-selinux test_21a+21b
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 8c1b62d09484161a40071c0aeecd1f1753d56870
Sebastien Buisson (sbuisson@ddn.com) uploaded a new patch: https://review.whamcloud.com/38034
Subject: LU-13156 tests: wait for nodemap update in sanity-selinux
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 164b0381dddbd3a81e614493d30b811f6297348a
Thanks to the testing carried out under https://review.whamcloud.com/37902, we can see that the problem hit with sanity-selinux test_21a and test_21b is due to primary MDS reboot/restart in prior tests launched in recovery-small, like test_23 or test_141.
Because of such primary MDS reboot/restart, we need to make sure that nodemaps are properly updated on all nodes before using them.
Oleg Drokin (green@whamcloud.com) merged in patch https://review.whamcloud.com/38034/
Subject: LU-13156 tests: wait for nodemap update in sanity-selinux
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: f1761cbe6b1243edd7a69c68c401d7285f7f3b38
Landed for 2.14
"Etienne AUJAMES <eaujames@ddn.com>" uploaded a new patch: https://review.whamcloud.com/c/fs/lustre-release/+/50400
Subject: LU-13156 tests: wait for nodemap update in sanity-selinux
Project: fs/lustre-release
Branch: b2_12
Current Patch Set: 1
Commit: dc79dedfea4dbd46f1f9fcf2ac0c1e9f9ff25a6b