I am going to fix it as soon as i will be able to test it, as far as it doesn't seem convenient to fix something i can't reproduce. For now i just brought the patch to the current state. >Четверг, 21 ноября 2019, 2:01 +03:00 от Vladislav Shpilevoy : > >Hm. So you are not going to fix the flaky error I mentioned >in the previous thread about this commit? > >Seems like it is also about 'conditions which need time to be >satisfied'. > >On 20/11/2019 23:47, Ilya Kosarev wrote: >> There were some pass conditions in quorum test which could take some >> time to be satisfied. Now they are wrapped using test_run:wait_cond to >> make the test stable. >> >> Part of #4586 >> --- >> test/replication/quorum.result | 30 +++++++++++++++++------------- >> test/replication/quorum.test.lua | 18 +++++++++--------- >> 2 files changed, 26 insertions(+), 22 deletions(-) >> >> diff --git a/test/replication/quorum.result b/test/replication/quorum.result >> index ff5fa0150..12604c8de 100644 >> --- a/test/replication/quorum.result >> +++ b/test/replication/quorum.result >> @@ -115,15 +115,15 @@ box.info.status -- running >> - running >> ... >> -- Check that the replica follows all masters. >> -box.info.id == 1 or box.info.replication[1].upstream.status == 'follow' >> +box.info.id == 1 or test_run:wait_cond(function() return box.info.replication[1].upstream.status == 'follow' end, 20) >> --- >> - true >> ... >> -box.info.id == 2 or box.info.replication[2].upstream.status == 'follow' >> +box.info.id == 2 or test_run:wait_cond(function() return box.info.replication[2].upstream.status == 'follow' end, 20) >> --- >> - true >> ... >> -box.info.id == 3 or box.info.replication[3].upstream.status == 'follow' >> +box.info.id == 3 or test_run:wait_cond(function() return box.info.replication[3].upstream.status == 'follow' end, 20) >> --- >> - true >> ... >> @@ -149,6 +149,10 @@ test_run:cmd('stop server quorum1') >> --- >> - true >> ... >> +test_run:wait_cond(function() return box.space.test.index.primary ~= nil end, 20) >> +--- >> +- true >> +... >> for i = 1, 100 do box.space.test:insert{i} end >> --- >> ... >> @@ -166,9 +170,9 @@ test_run:cmd('switch quorum1') >> --- >> - true >> ... >> -box.space.test:count() -- 100 >> +test_run:wait_cond(function() return box.space.test:count() == 100 end, 20) >> --- >> -- 100 >> +- true >> ... >> -- Rebootstrap one node of the cluster and check that others follow. >> -- Note, due to ERRINJ_RELAY_TIMEOUT there is a substantial delay >> @@ -197,9 +201,9 @@ test_run:cmd('switch quorum1') >> - true >> ... >> test_run:cmd('restart server quorum1 with cleanup=1, args="0.1 0.5"') >> -box.space.test:count() -- 100 >> +test_run:wait_cond(function() return box.space.test:count() == 100 end, 20) >> --- >> -- 100 >> +- true >> ... >> -- The rebootstrapped replica will be assigned id = 4, >> -- because ids 1..3 are busy. >> @@ -207,11 +211,9 @@ test_run:cmd('switch quorum2') >> --- >> - true >> ... >> -fiber = require('fiber') >> ---- >> -... >> -while box.info.replication[4].upstream.status ~= 'follow' do fiber.sleep(0.001) end >> +test_run:wait_cond(function() return box.info.replication[4].upstream.status == 'follow' end, 20) >> --- >> +- true >> ... >> box.info.replication[4].upstream.status >> --- >> @@ -221,11 +223,13 @@ test_run:cmd('switch quorum3') >> --- >> - true >> ... >> -fiber = require('fiber') >> +test_run:wait_cond(function() return box.info.replication ~= nil end, 20) >> --- >> +- true >> ... >> -while box.info.replication[4].upstream.status ~= 'follow' do fiber.sleep(0.001) end >> +test_run:wait_cond(function() return box.info.replication[4].upstream.status == 'follow' end, 20) >> --- >> +- true >> ... >> box.info.replication[4].upstream.status >> --- >> diff --git a/test/replication/quorum.test.lua b/test/replication/quorum.test.lua >> index 98febb367..be23200d3 100644 >> --- a/test/replication/quorum.test.lua >> +++ b/test/replication/quorum.test.lua >> @@ -47,9 +47,9 @@ box.info.ro -- false >> box.info.status -- running >> >> -- Check that the replica follows all masters. >> -box.info.id == 1 or box.info.replication[1].upstream.status == 'follow' >> -box.info.id == 2 or box.info.replication[2].upstream.status == 'follow' >> -box.info.id == 3 or box.info.replication[3].upstream.status == 'follow' >> +box.info.id == 1 or test_run:wait_cond(function() return box.info.replication[1].upstream.status == 'follow' end, 20) >> +box.info.id == 2 or test_run:wait_cond(function() return box.info.replication[2].upstream.status == 'follow' end, 20) >> +box.info.id == 3 or test_run:wait_cond(function() return box.info.replication[3].upstream.status == 'follow' end, 20) >> >> -- Check that box.cfg() doesn't return until the instance >> -- catches up with all configured replicas. >> @@ -59,13 +59,14 @@ test_run:cmd('switch quorum2') >> box.error.injection.set("ERRINJ_RELAY_TIMEOUT", 0.001) >> test_run:cmd('stop server quorum1') >> >> +test_run:wait_cond(function() return box.space.test.index.primary ~= nil end, 20) >> for i = 1, 100 do box.space.test:insert{i} end >> fiber = require('fiber') >> fiber.sleep(0.1) >> >> test_run:cmd('start server quorum1 with args="0.1 0.5"') >> test_run:cmd('switch quorum1') >> -box.space.test:count() -- 100 >> +test_run:wait_cond(function() return box.space.test:count() == 100 end, 20) >> >> -- Rebootstrap one node of the cluster and check that others follow. >> -- Note, due to ERRINJ_RELAY_TIMEOUT there is a substantial delay >> @@ -81,17 +82,16 @@ box.snapshot() >> test_run:cmd('switch quorum1') >> test_run:cmd('restart server quorum1 with cleanup=1, args="0.1 0.5"') >> >> -box.space.test:count() -- 100 >> +test_run:wait_cond(function() return box.space.test:count() == 100 end, 20) >> >> -- The rebootstrapped replica will be assigned id = 4, >> -- because ids 1..3 are busy. >> test_run:cmd('switch quorum2') >> -fiber = require('fiber') >> -while box.info.replication[4].upstream.status ~= 'follow' do fiber.sleep(0.001) end >> +test_run:wait_cond(function() return box.info.replication[4].upstream.status == 'follow' end, 20) >> box.info.replication[4].upstream.status >> test_run:cmd('switch quorum3') >> -fiber = require('fiber') >> -while box.info.replication[4].upstream.status ~= 'follow' do fiber.sleep(0.001) end >> +test_run:wait_cond(function() return box.info.replication ~= nil end, 20) >> +test_run:wait_cond(function() return box.info.replication[4].upstream.status == 'follow' end, 20) >> box.info.replication[4].upstream.status >> >> -- Cleanup. >> -- Ilya Kosarev