Bug #32991 mysqlimport --use-threads test fails on solaris

Description:
The following code in mysqldump.test sometimes causes failures:
------------------------------------------------------------
#
# Added for use-thread option
#
create table t1 (a text , b text);
create table t2 (a text , b text);
insert t1 values ("Duck, Duck", "goose");
insert t1 values ("Duck, Duck", "pidgeon");
insert t2 values ("We the people", "in order to perform");
insert t2 values ("a more perfect", "union");
select * from t1;
select * from t2;
--exec $MYSQL_DUMP --tab=$MYSQLTEST_VARDIR/tmp/ test
--exec $MYSQL test < $MYSQLTEST_VARDIR/tmp/t1.sql
--exec $MYSQL test < $MYSQLTEST_VARDIR/tmp/t2.sql
# The first load tests the pausing code
--exec $MYSQL_IMPORT --use-threads=1 test $MYSQLTEST_VARDIR/tmp/t1.txt $MYSQLTEST_VARDIR/tmp/t2.txt
# Now we test with multiple threads!
--exec $MYSQL_IMPORT --silent --use-threads=5 test $MYSQLTEST_VARDIR/tmp/t1.txt $MYSQLTEST_VARDIR/tmp/t2.txt
select * from t1;
select * from t2;
# Now we test with multiple threads, but less threads than files.
create table words(a varchar(255));
create table words2(b varchar(255));
--exec $MYSQL_IMPORT --silent --use-threads=2 test $MYSQLTEST_VARDIR/tmp/t1.txt $MYSQLTEST_VARDIR/tmp/t2.txt $MYSQLTEST_VARDIR/std_data_ln/words.dat $MYSQLTEST_VARDIR/std_data_ln/words2.dat
select * from t1;
select * from t2;
select * from words;
select * from words2;
------------------------------------------------------------

The failures look like:

------------------------------------------------------------
main.mysqldump                 [ fail ]

mysqldump: Couldn't find table: "non_existing"
mysqldump: Got error: 1356: View 'test.v1' references invalid table(s) or column(s) or function(s) or definer/invoker of view lack rights to use them when using LOCK TABLES
mysqldump: Couldn't execute 'SHOW FIELDS FROM `v1`': View 'test.v1' references invalid table(s) or column(s) or function(s) or definer/invoker of view lack rights to use them (1356)
mysqldump: user2 has insufficent privileges to SHOW CREATE PROCEDURE `sp1`!
--- /export/home/pushbuild/pb/mysql-5.2-maint/37/mysql-6.0.4-alpha-pb37/mysql-test/r/mysqldump.result	Sat Dec  1 16:04:04 2007
+++ /export/home/pushbuild/pb/mysql-5.2-maint/37/mysql-6.0.4-alpha-pb37/mysql-test/r/mysqldump.reject	Sun Dec  2 15:01:23 2007
@@ -4019,7 +4019,7 @@
 aberrant
 aberration
 drop table words;
-mysql-import: Error: 1146, Table 'test.words' doesn't exist, when using table: words
+mysql-import: Error: 2002 Can't connect to local MySQL server through socket '/tmp/pbtmp-n_mix-100/master.sock' (146)
 drop table t1;
 drop table t2;
 drop table words2;

mysqltest: Result content mismatch

Stopping All Servers
Restoring snapshot of databases
Resuming Tests
------------------------------------------------------------

or

------------------------------------------------------------
main.mysqldump                 [ fail ]

mysqldump: Couldn't find table: "non_existing"
mysqldump: Got error: 1356: View 'test.v1' references invalid table(s) or column(s) or function(s) or definer/invoker of view lack rights to use them when using LOCK TABLES
mysqldump: Couldn't execute 'SHOW FIELDS FROM `v1`': View 'test.v1' references invalid table(s) or column(s) or function(s) or definer/invoker of view lack rights to use them (1356)
mysqldump: user2 has insufficent privileges to SHOW CREATE PROCEDURE `sp1`!
/export/home/pushbuild/pb/mysql-5.1-runtime/586/mysql-5.1.23-rc-pb586/client/mysqlimport: Error: 2002 Can't connect to local MySQL server through socket '/tmp/pbtmp-n_row-5/master.sock' (146)
mysqltest: At line 1636: command "$MYSQL_IMPORT --silent --use-threads=2 test $MYSQLTEST_VARDIR/tmp/t1.txt $MYSQLTEST_VARDIR/tmp/t2.txt $MYSQLTEST_VARDIR/std_data_ln/words.dat $MYSQLTEST_VARDIR/std_data_ln/words2.dat" failed

The result from queries just before the failure was:
< snip >
a	b
We the people	in order to perform
a more perfect	union
test.t1: Records: 2  Deleted: 0  Skipped: 0  Warnings: 0
test.t2: Records: 2  Deleted: 0  Skipped: 0  Warnings: 0
select * from t1;
a	b
Duck, Duck	goose
Duck, Duck	pidgeon
Duck, Duck	goose
Duck, Duck	pidgeon
select * from t2;
a	b
We the people	in order to perform
a more perfect	union
We the people	in order to perform
a more perfect	union
create table words(a varchar(255));
create table words2(b varchar(255));
exec of '/export/home/pushbuild/pb/mysql-5.1-runtime/586/mysql-5.1.23-rc-pb586/client/mysqlimport -uroot --debug-check --port=10050 --socket=/tmp/pbtmp-n_row-5/master.sock --password= --silent --use-threads=2 test /tmp/var-n_row-5/tmp/t1.txt /tmp/var-n_row-5/tmp/t2.txt /tmp/var-n_row-5/std_data_ln/words.dat /tmp/var-n_row-5/std_data_ln/words2.dat' failed, error: 256, status: 1, errno: 29

More results from queries before failure can be found in /tmp/var-n_row-5/log/mysqldump.log

Stopping All Servers
Restoring snapshot of databases
Resuming Tests
------------------------------------------------------------

How to repeat:
https://intranet.mysql.com/secure/pushbuild/xref.pl?testname=main.mysqldump
https://intranet.mysql.com/secure/pushbuild/xref.pl?plat=not+%25win2003%25&testname=main.m...

Experimented a little more with this and it is easily reproduced on sol10-sparc-a.

The problem seems to be that mysqlimport fsils to connect when run with --n um-threads>1, is it really built with the thread safe library?

To reproduce, build mysql-5.1 on sol10-sparc-a. Start the server with
./mtr mysqldump --start-and-exit

Create and fill tables:

mysql> create table t1 (a text , b text);
mysql> create table t2 (a text , b text);
mysql> insert t1 values ("Duck, Duck", "goose");
mysql> insert t1 values ("Duck, Duck", "pidgeon");
mysql> insert t2 values ("We the people", "in order to perform");
mysql> insert t2 values ("a more perfect", "union");
mysql> select * from t1;
mysql> select * from t2;

Run mysqldump to create t1.txt and t2. txt in var/tmp
../client/mysqldump -S var/tm p/master.socl k--tab=var/tmp/ test

Run mysqlimport continuously in a while loop.
$> while mysqlimport.sh ; do date; done

It will fail and report that mysqlimport couldn't connect to mysqld through socket

mysqlimport.sh would look something like:
../client/mysqlimport -uroot --debug-check --port=10170 --socket=/tmp/MfLgwlfAyM/master.sock --password= --silent --use-threads=2 test /export/home/mysqldev/users/magnus/mysql-5.1.24-rc-pb472/mysql-test/var/tmp/t1.txt /export/home/mysqldev/users/magnus/mysql-5.1.24-rc-pb472/mysql-test/var/tmp/t2.txt /export/home/mysqldev/users/magnus/mysql-5.1.24-rc-pb472/mysql-test/var/std_data_ln/words.dat /export/home/mysqldev/users/magnus/mysql-5.1.24-rc-pb472/mysql-test/var/std_data_ln/words2.dat $1

Posible problems:
1. Not using thread safe library. All clients in client/ is compiled with DUNDEF_THREADS_HACK, remove that defined and let each client include what they need to get either thread safe or non thread safe code.
2. The mysqlimport need a small retry loop when connecting - not so good, the clients should connect every time.

"All clients in client/ is compiled with
DUNDEF_THREADS_HACK," and then we go though hoops in order to get thread safety in the one that needs it.

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/42228

ChangeSet@1.2548, 2008-02-13 17:39:23+01:00, sven@riska.(none) +2 -0
  BUG#32991: Races in mysqldump.test (or mysqldump.test fails sporadically)
  This is *not* a fix to the bug. I'm only disabling the failing part of
  mysqldump.test until the bug is fixed. Whoever fixes it, please re-enable
  the test.

The submitted patch does *not* fix the bug. It only disables the test case until the bug is fixed, so that not everyone gets errors in pushbuild. Whoever fixes this bug, please re-enable the test.

The error is in mysqlimport, and happens when the --use-threads flag is used. I looked superficially at it (checked global variables used from worker_thread() in client/mysqlimport.c), and could not see anything strange. (Except that the global 'exitcode' variable is used without a mutex, but it should not cause this sort of error).

The error log on sol10-sparc-a says "errno=23", which on this machine means "illegal seek". I'm just guessing, but could printf() be non-thread-safe on this platform?

Pushed into 5.1.24-rc

Pushed into 6.0.5-alpha

Considering that there's a comment about the -DUNDEF_THREADS_HACK saying it a fix for MIT-pthreads, it can probably just be removed. Otherwise the same CFLAGS used for mysqlslap should be used for mysqlimport.

Unset UNDEF_THREADS_HACK for mysqlimport, because it uses threads

Attachment: bug32991.patch (text/plain), 426 bytes.

I can't reproduce the problem on sol10-sparc-a, but the attached patch will compile mysqlimport the same way as mysqlslap, to avoid the UNDEF_THREADS_HACK hack.

Probably bug #37789 is related.

Looks ok; I agree it would be nice to get rid of UNDEF_THREADS_HACK entirely, but maybe needs more research.  This patch is simple.

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/73853

2894 Jim Winstead	2009-05-12
      mysqlimport was not always compiled correctly to allow thread support,
      required for the --use-threads option. (Bug #32991)
      modified:
        client/Makefile.am
        mysql-test/r/mysqldump.result
        mysql-test/t/mysqldump.test

Pushed to 5.1-bugteam and mysql-pe.

Pushed into 5.4.4-alpha (revid:alik@sun.com-20090804194615-h40sa098mx4z49qg) (version source revid:jimw@mysql.com-20090713193847-xj6gbgm2a1zwwanv) (merge vers: 5.4.4-alpha) (pib:11)

Pushed into 5.1.38 (revid:davi.arnaut@sun.com-20090804204317-ggodqkik7de6nfpz) (version source revid:davi.arnaut@sun.com-20090804204317-ggodqkik7de6nfpz) (merge vers: 5.1.38) (pib:11)

Noted in 5.1.38, 5.4.4 changelogs.

mysqlimport was not always compiled correctly to enable thread
support, which is required for the --use-threads option.

Noted in 5.4.2 changelog because next 5.4 version will be 5.4.2 and not 5.4.4.

Ignore previous comment about 5.4.2.

Pushed into 5.1.39-ndb-6.3.28 (revid:jonas@mysql.com-20091001055605-ap2kiaarr7p40mmv) (version source revid:jonas@mysql.com-20091001055605-ap2kiaarr7p40mmv) (merge vers: 5.1.39-ndb-6.3.28) (pib:11)

Pushed into 5.1.39-ndb-7.0.9 (revid:jonas@mysql.com-20091001072547-kv17uu06hfjhgjay) (version source revid:jonas@mysql.com-20091001071652-irejtnumzbpsbgk2) (merge vers: 5.1.39-ndb-7.0.9) (pib:11)

Pushed into 5.1.39-ndb-7.1.0 (revid:jonas@mysql.com-20091001123013-g9ob2tsyctpw6zs0) (version source revid:jonas@mysql.com-20091001123013-g9ob2tsyctpw6zs0) (merge vers: 5.1.39-ndb-7.1.0) (pib:11)

Pushed into 5.1.39-ndb-6.2.19 (revid:jonas@mysql.com-20091005103850-dwij2dojwpvf5hi6) (version source revid:jonas@mysql.com-20090930185117-bhud4ek1y0hsj1nv) (merge vers: 5.1.39-ndb-6.2.19) (pib:11)

The 5.4 fix has been pushed to 5.4.2.