From d23e8e06a184bb7ed4f97afb4d4c0cc696788512 Mon Sep 17 00:00:00 2001 From: Matthias Leich Date: Wed, 13 Aug 2008 21:44:54 +0200 Subject: [PATCH] Fix for Bug#37853 Test "funcs_1.processlist_val_ps" fails in various ways + corrections of logic in poll routines + minor improvements --- .../funcs_1/datadict/processlist_val.inc | 612 +++++++++++------- .../funcs_1/r/processlist_val_no_prot.result | 298 ++++++--- .../suite/funcs_1/r/processlist_val_ps.result | 298 ++++++--- 3 files changed, 784 insertions(+), 424 deletions(-) diff --git a/mysql-test/suite/funcs_1/datadict/processlist_val.inc b/mysql-test/suite/funcs_1/datadict/processlist_val.inc index 16c3dc71ff1..c34fb626bcd 100644 --- a/mysql-test/suite/funcs_1/datadict/processlist_val.inc +++ b/mysql-test/suite/funcs_1/datadict/processlist_val.inc @@ -1,313 +1,465 @@ -########### suite/funcs_1/datadict/processlist_val.inc ################# -# # -# Testing of values within INFORMATION_SCHEMA.PROCESSLIST # -# # -# Ensure that the values fit to the current event of the connection # -# and especially that they change if a connection does nothing or # -# runs some SQL. # -# Examples: # -# - change the default database # -# - send some time no SQL command to the server # -# - send a long running query # -# # -# Note(mleich): # -# 1. Please inform me if this test fails because of timing problems. # -# I tried to avoid instabilities but the values within the column # -# TIME are very sensible to fluctuations of the machine load. # -# I had to unify some TIME values with "--replace_result" in cases # -# where they are too unstable. # -# 2. Storage engine variants of this test do not make sense. # -# - I_S tables use the MEMORY storage engine whenever possible. # -# - There are some I_S table which need column data types which # -# are not supported by MEMORY. Example: LONGTEXT/BLOB # -# MyISAM will be used for such tables. # -# The column PROCESSLIST.INFO is of data type LONGTEXT # -# ----> MyISAM # -# - There is no impact of the GLOBAL(server) or SESSION default # -# storage engine setting on the engine used for I_S tables. # -# That means we cannot get NDB or InnoDB instead. # -# 3. The SHOW (FULL) PROCESSLIST command are for comparison. # -# The main test target is INFORMATION_SCHEMA.PROCESSLIST ! # -# # -# Creation: # -# 2007-08-09 mleich Implement this test as part of # -# WL#3982 Test information_schema.processlist # -# 2008-01-05 HHunger Changed time test and long statement test. # -# It could happen that the long statement test # -# fail due to timing problems. # -# 2008-04-14 pcrews Changed --replace_columns to include host # -# this is due to failure occurring on Windows # -# Win results included port number as well causing # -# test failures # -######################################################################## +########### suite/funcs_1/datadict/processlist_val.inc ######################### +# # +# Testing of values within INFORMATION_SCHEMA.PROCESSLIST # +# # +# Ensure that the values fit to the current state of the connection and # +# especially that they change if a connection does nothing or runs some SQL. # +# runs some SQL. # +# Examples: # +# - change the default database # +# - send some period of time no SQL command to the server # +# - send a long running query # +# # +# Note(mleich): # +# 1. Please inform me if this test fails because of timing problems. # +# 2. Storage engine variants of this test do not make sense. # +# - I_S tables use the MEMORY storage engine whenever possible. # +# - There are some I_S tables which need column data types which are not # +# supported by MEMORY. Example: LONGTEXT/BLOB # +# MyISAM will be used for such tables. # +# The column PROCESSLIST.INFO is of data type LONGTEXT ----> MyISAM # +# - There is no impact of the GLOBAL(server) or SESSION default storage # +# engine setting on the engine used for I_S tables. # +# That means we cannot get NDB or InnoDB instead. # +# 3. The SHOW (FULL) PROCESSLIST command are for comparison. # +# The main test target is INFORMATION_SCHEMA.PROCESSLIST ! # +# 4. Attention: # +# The values of the PROCESSLIST columns HOST and TIME tend to cause # +# problems and therefore their printing has to be suppressed. # +# Examples of the exact values: # +# HOST: 'localhost' (UNIX derivates) # +# 'localhost:' (WINDOWS) # +# TIME: In many cases within this test 0 seconds but if the testing box is # +# overloaded we might get up to 2 seconds. # +# Solution: # +# --replace_column 1 3 6