Description of problem: I found that tests/basic/ec/ec-1468261.t is timing out because of the change in heal-timeout to 10 minutes from 1 minute in this patch. I changed it to 1 minute in this patch. But after that I am running into another issue because of the recent change in kill_brick() implementation. The test at line 38 creates a new file and immediately kills two bricks and checks that trusted.ec.dirty xattr is non-zero for the bricks. But what ends up happening is that even before the brick is down, the post-op happens, leading to failure of the test. [2018-08-16 08:14:38.384004]:++++++++++ G_LOG:tests/basic/ec/ec-1468261.t: TEST: 38 touch /mnt/glusterfs/0/test_dir/new_file ++++++++++ [2018-08-16 08:14:38.576104]:++++++++++ G_LOG:tests/basic/ec/ec-1468261.t: TEST: 39 kill_brick patchy localhost.localdomain /d/backends/patchy0 ++++++++++ [2018-08-16 08:14:39.723034] I [ec-common.c:2409:ec_unlock_now] 0-UNLOCKING: cd87ef5f-a7cc-4dc0-9545-9fdc6a66d362 ****** Post-op and Unlock on the directory happened in the line above after that the disconnect happened below ****** [2018-08-16 08:14:39.766402] W [socket.c:731:__socket_rwv] 0-patchy-client-0: readv on 127.0.0.1:49152 failed (No data available) Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Still relevant?
Not observed anymore.