Igor, Thanks for the review, I've changed the commit message as you suggested. >Среда, 15 января 2020, 2:25 +03:00 от Igor Munkin : > >Sasha, > >Thanks, the patch LGTM, though as discussed offline I see no relation to >ASAN. So I propose to reword the commit message as the following: >| All local connection timeout settings not related to the testing >| scenario are removed within this change. Instead of removed values the >| default one from src/box/lua/load_cfg.lua will be used. The approach >| with a single default value helps to avoid flaky test results >| regarding different timeout values and makes the future maintainence >| easier. >| >| The change is required for LSAN and ASAN testing machinery since it >| introduces a little overhead and cause failures for tests with >| excessively strict time limits. >| >| Needed for #2058 > >Feel free to adjust it on your own way. > >On 21.11.19, Alexander V. Tikhonov wrote: >> Removed all local connection timeout setups, where it is not the part >> of the testing scenario. Instead of it the default value will be used. >> Also it is more comfortable to change the single default value instead >> of the change this value at each of the test scenarios, it will help >> to avoid of flaky results in tests beacuse of different timeout vlaues. > >Typo: s/vlaues/values/. >Typo: s/beacuse/because/. > >> The most of tests used the replication connection timeout is equal to >> 0.5 second, while defualt value is set at sources file: > >Typo: s/defualt/default/. > >> src/box/lua/load_cfg.lua >> to the value of 30 seconds: >> replication_connect_timeout = 30 >> which is quite enough for all of the tests that don't need any special >> values for it. >> >> Part of #2058 >> --- > > > >> -- >> 2.17.1 >> > >-- >Best regards, >IM -- Alexander Tikhonov