Details
-
Bug
-
Status: Closed (View Workflow)
-
Minor
-
Resolution: Won't Fix
-
10.1(EOL)
-
None
Description
main.status w2 [ fail ]
|
Test ended at 2015-07-20 18:41:05
|
|
|
CURRENT_TEST: main.status
|
--- /usr/local/mariadb-10.0.20-linux-x86_64/mysql-test/r/status.result 2015-07-20 14:41:12.000000000 +0300
|
+++ /usr/local/mariadb-10.0.20-linux-x86_64/mysql-test/r/status.reject 2015-07-20 18:41:05.098326356 +0300
|
@@ -182,7 +182,7 @@
|
Variable_name Value
|
Com_show_status 8
|
rnd_diff tmp_table_diff
|
-28 8
|
+26 8
|
flush status;
|
show status like 'Com%function';
|
Variable_name Value
|
|
|
mysqltest: Result content mismatch
|
Fails from time to time on different builders. I could not reproduce it even by running the test 1000 times. Sequence of tests should not be important, since the server gets restarted before the test.
Attachments
Issue Links
- blocks
-
MDEV-7069 Fix buildbot failures in main server trees
-
- Stalled
-
- relates to
-
-
- Closed
-
Activity
A new flavor of failure occurred in buildbot at least once:
http://buildbot.askmonty.org/buildbot/builders/winx64-debug/builds/237/steps/test/logs/stdio
main.status w4 [ fail ]
|
Test ended at 2016-06-03 12:22:51
|
|
|
CURRENT_TEST: main.status
|
--- D:/winx64-debug/build/src/mysql-test/r/status.result 2016-06-03 11:47:05.223039600 +0000
|
+++ D:\winx64-debug\build\src\mysql-test\r\status.reject 2016-06-03 12:22:51.397420800 +0000
|
@@ -242,9 +242,9 @@
|
CALL p1();
|
1
|
1
|
-SELECT 9;
|
-9
|
-9
|
+SELECT 10;
|
+10
|
+10
|
DROP PROCEDURE p1;
|
DROP FUNCTION f1;
|
flush status;
|
|
|
mysqltest: Result length mismatch
|
For this latter failure I have added log dump in case of failure and it appears to be caused by earlier testcase disconnects completing asynchronously during this logging-enabled window. The testcase tries to wait for some of the disconnects to complete by waiting for the threads to disappear from processlist, but 1) this is not universal for all connections; 2) the threads disappear from processlist early in the disconnect process.
I have changed the testcase to use wait_until_count_sessions.inc (Threads_connected decrement happens very late in the disconnect): https://github.com/percona/percona-server/pull/1679. The diff contains semi-related changes of reverting that previously added dump
The problem has not happened in buildbot on 10.0 since July 20, 2015 and on 10.1 since August 2015. Before that, it was happening ~2-4 times a month, so I think we can safely assume it disappeared from the trees. It's possible it was somehow related to (and fixed with)
MDEV-8301, the timing fits.