- developers - lists.mariadb.org

[Maria-developers] [MDEV-7362] ANALYZE TABLES crash with table-independent-statistics gathering
by Vicențiu Ciorbaru 17 Jan '15

17 Jan '15

Hi Sergei! I've implemented a fix for the bug described in MDEV-7362, that also fixes MDEV-7380. The test case also covers MDEV-7380. With the following fix,the I've spent a bit of time trying to figure out the architecture of things there. I am not confident my fix is complete, or if this is the way we want to go about fixing it. The fix does what we discussed over IRC, that we are not interested in computing statistics for a FULLTEXT index. What I think would be the root cause of the problem is calling ha_index_init() on an index that does not support this. By calling table->file->ha_index_init(), the table->field array gets corrupted and that ultimately leads to a crash in the function using it afterwards. Then again, this module is completely new to me, so any comments are appreciated. Here is the diff of the first changes. diff --git a/mysql-test/t/mdev-7362.test b/mysql-test/t/mdev-7362.test new file mode 100644 index 0000000..6551893 --- /dev/null +++ b/mysql-test/t/mdev-7362.test @@ -0,0 +1,14 @@ +CREATE TABLE t1 (a longtext, FULLTEXT KEY (`a`)) ENGINE=MyISAM; +insert into t1 values (unhex('3E0D0A4141414142334E7A6143317963324541414141424977414141674541726D')); +analyze table t1 persistent for all; +drop table t1; + +CREATE TABLE t1 (a longtext, FULLTEXT KEY (`a`)) ENGINE=MyISAM; +insert into t1 values (unhex('3E0D0A4141414142334E7A6143317963324541414141424977414141674541')); +analyze table t1 persistent for all; +drop table t1; + +CREATE TABLE t1 (f longtext NOT NULL, FULLTEXT KEY f (f)) ENGINE=MyISAM; +INSERT INTO t1 VALUES (REPEAT('a',27)),('foo'); +ANALYZE TABLE t1 PERSISTENT FOR ALL; +drop table t1; diff --git a/sql/sql_statistics.cc b/sql/sql_statistics.cc index d368145..2c1cccf 100644 --- a/sql/sql_statistics.cc +++ b/sql/sql_statistics.cc @@ -2367,6 +2367,9 @@ int collect_statistics_for_index(THD *thd, TABLE *table, uint index) DBUG_RETURN(rc); } + if (key_info->flags & HA_FULLTEXT) // skip if FULLTEXT index + DBUG_RETURN(rc); + table->key_read= 1; table->file->extra(HA_EXTRA_KEYREAD); Regards, Vicențiu

3 5

[Maria-developers] MDEV-7440 Build fails in libgroonga on Solaris
by Alexander Barkov 16 Jan '15

16 Jan '15

Hello Kentoku, Can you please have a look into this problem? https://mariadb.atlassian.net/browse/MDEV-7440 Can you fix please? Thanks.

1 0

Re: [Maria-developers] [Commits] Rev 4398: MDEV-7368 : SLES: Failed to start mysql.service: Unit
by Sergei Golubchik 15 Jan '15

15 Jan '15

Hi, Nirbhay! On Dec 23, Nirbhay Choubey wrote: > At lp:~maria-captains/maria/5.5 > > ------------------------------------------------------------ > revno: 4398 > revision-id: nirbhay(a)mariadb.com-20141224033207-8lh5i8d84qsjey79 > parent: elenst(a)montyprogram.com-20141223172123-pkfhjnv7oy1suxpa > committer: Nirbhay Choubey <nirbhay(a)mariadb.com> > branch nick: b7368-5.5 > timestamp: Tue 2014-12-23 22:32:07 -0500 > message: > MDEV-7368 : SLES: Failed to start mysql.service: Unit > mysql.service failed to load > > Added 'systemctl daemon-reload' command in postin and > postun rpm scripts. > === modified file 'support-files/rpm/server-postin.sh' > --- a/support-files/rpm/server-postin.sh 2013-11-20 11:18:46 +0000 > +++ b/support-files/rpm/server-postin.sh 2014-12-24 03:32:07 +0000 > @@ -1,6 +1,10 @@ > > # Make MySQL start/shutdown automatically when the machine does it. > if [ $1 = 1 ] ; then > + if [ -x /usr/bin/systemctl ] ; then > + /usr/bin/systemctl daemon-reload >/dev/null 2>&1 > + fi > + looks ok > if [ -x /sbin/chkconfig ] ; then > /sbin/chkconfig --add mysql > fi > @@ -51,30 +55,31 @@ > SETARGETDIR=/etc/selinux/targeted/src/policy > SEDOMPROG=$SETARGETDIR/domains/program > SECONPROG=$SETARGETDIR/file_contexts/program > +if [ -f /etc/redhat-release ] ; then > + if grep -q "Red Hat Enterprise Linux .. release 4" /etc/redhat-release \ > + || grep -q "CentOS release 4" /etc/redhat-release ; then I'd rather written if grep -Eqs '(Red Hat Enterprise Linux ..|CentOS) release 4' /etc/redhat-release; then or (which is more portable, but perhaps unnecessary so for an rpm postin script) if grep '$Red Hat Enterprise Linux ..\|CentOS$ release 4' /etc/redhat-release >/dev/null 2>&1; then but if you'd like you can push your variant. Regards, Sergei

2 1

Re: [Maria-developers] [Commits] Rev 4401: MDEV-7448 - mtr may leave stale mysqld in lp:maria/5.5
by Sergei Golubchik 14 Jan '15

14 Jan '15

Hi, Sergey! On Jan 14, Sergey Vojtovich wrote: > ------------------------------------------------------------ > revno: 4401 > revision-id: svoj(a)mariadb.org-20150114135038-jss2vvpkhihysfmj > parent: sergii(a)pisem.net-20150110130746-xlrofqddt4j8xkw6 > committer: Sergey Vojtovich <svoj(a)mariadb.org> > branch nick: 5.5 > timestamp: Wed 2015-01-14 17:50:38 +0400 > message: > MDEV-7448 - mtr may leave stale mysqld > > Let mtr handle SIGHUP the same way as SIGINT. This solves stale mysqld processes > left after broken/closed ssh connection. Looks ok. But please make sure that all tests pass before pushing it. mysqld uses SIGHUP to print various information to the error log. If there's a test that actually tests that by sending SIGHUP - your change might've broken it. Regards, Sergei

1 0

[Maria-developers] MySQL bug #68557 "sjis changes in 5.5 break date parsing" fixed in MariaDB >= 5.5.38
by Hartmut Holzgraefe 13 Jan '15

13 Jan '15

Any idea how that happened, as I can't find anything that seems to be related in the 5.5.38 changelog? http://bugs.mysql.com/bug.php?id=68557 -- Hartmut Holzgraefe, Principal Support Engineer (EMEA) MariaDB Corporation | http://www.mariadb.com/

2 1

[Maria-developers] Buildbot slow upload
by Kristian Nielsen 13 Jan '15

13 Jan '15

Hi Serg, Daniel, Some time ago there was discussion of slow Buildbot upload speed (sorry, don't have the original mail thread to reply to). I found some new information on this. Yesterday I had to do a custom build, but was not able to because my tarbake upload kept getting stuck. So I had to investigate. Turns out the real problem was that hasky:/media/backup/archive/pack was full. And apparently, the error handling in Buildbot upload step is broken, so the error is not caught properly but just causes the upload to hang. I have noticed this before, that upload gets stuck. I wonder if full disk on hasky has caused this before. Daniel: Do we have something that removes old builds from /archive/pack? I did not see anything in root or buildbot crontabs? With respect to slow uploads, while trying various things, one thing I did was increase the upload blocksize in Buildbot config. Now I see tarbake upload (50MB) taking just 1 minute, which seems fair (probably it can be more if the network or buildbot is busy). So I'm wondering if the slow upload that was seen was in fact a combination of too low default blocksize, and uploads occasionally getting _completely_ stuck? It's of course still fine to move to using rsync, as Serg proposed earlier; just wanted to mention what I found. And in either case, we still need a solution to prevent /archive/pack from running full in the future... - Kristian.

2 1

[Maria-developers] MDEV-7288 USER/ROLE: CREATE OR REPLACE, CREATE IF NOT EXISTS, DROP IF EXISTS
by Alexander Barkov 09 Jan '15

09 Jan '15

Hi Sergei, Please have a look into the patch for MDEV-7288. It consists of the USER/ROLE related part of the big GSoC patch from Sriram that you reviewed earlier, with review suggestions addressed. You earlier had the following suggestions/questions: >> diff --git a/sql/sql_acl.cc b/sql/sql_acl.cc >> index 5f077c2..1e156be 100644 >> --- a/sql/sql_acl.cc >> +++ b/sql/sql_acl.cc >> ... >> @@ -9215,10 +9252,26 @@ bool mysql_create_user(THD *thd, List <LEX_USER> &lis >> */ >> if (handle_grant_data(tables, 0, user_name, NULL)) >> { >> - append_user(thd, &wrong_users, user_name); >> - >> - result= TRUE; >> - continue; >> + if (thd->lex->is_create_or_replace()) >> + { >> + if (handle_grant_data(tables, 1, user_name, NULL) <= 0) >> + { >> + append_user(thd, &wrong_users, user_name); > > I don't understand that at all. If user creation failed, you try to drop > it, but don't create a user again, after the old one was dropped. How > could it work? > Yeah, it looks confusing, because handle_grant_data() can do different things depending on the arguments. The call for handle_grant_data(tables, 0, user_name, NULL) does not create any users. It only checks if the user already exists. The call for handle_grant_data(tables, 1, user_name, NULL) drops the user. And the call for replace_user_table() after this block is the one that actually creates the user. I have added some comments in the new code, to make it clearer. >>> @@ -9268,7 +9321,7 @@ bool mysql_create_user(THD *thd, List <LEX_USER> &list, >>> (handle_as_role) ? "CREATE ROLE" : "CREATE USER", >>> wrong_users.c_ptr_safe()); >>> >>> - if (some_users_created) >>> + if (some_users_created || !result) >> >> What is that for? To binlog CREATE IF NOT EXISTS? Yes, that was the reason. I renamed "some_users_created" to "binlog" and changed the logic slightly to set "binlog" to true under the block checking lex->create_info.if_not_exists(), instead of this additional test for "result". So now it should be easier to read: if (binlog) result |= write_bin_log(thd, FALSE, thd->query(), thd->query_length()); Thanks.

3 3

[Maria-developers] prefix matching for command-line options - how to disable?
by Sergei Golubchik 06 Jan '15

06 Jan '15

Hi, We've got a bit of a problem with command-line option prefixes in the test suite. As you know, in MariaDB a command line option (the same applies to my.cnf too) doesn't need to be specified completely, it's sufficient to use an unambigous prefix of it. For example, 'net-retry-count=5' in my.cnf can be abbreviated as 'net-ret=5'. But not as 'net-re=5', the latter will fail with an error mysqld: ambiguous option '--net-re' (net_read_timeout, net_retry_count) This is the historical behavior of MySQL since about 2002. But it creates problems for plugins. InnoDB is particularly badly affected - it has lots of I_S plugins that are prefixes of other I_S plugins. For example: INNODB_BUFFER_PAGE INNODB_BUFFER_PAGE_LRU ... INNODB_CMP INNODB_CMPMEM INNODB_CMPMEM_RESET INNODB_CMP_PER_INDEX INNODB_CMP_PER_INDEX_RESET INNODB_CMP_RESET ... INNODB_SYS_FOREIGN INNODB_SYS_FOREIGN_COLS ... INNODB_SYS_TABLES INNODB_SYS_TABLESPACES INNODB_SYS_TABLESTATS also, it has plugins that match other plugin variables, like INNODB_LOCKS and innodb_locks_unsafe_for_binlog. As a result one cannot see or predict what these options will do. For example --enable-innodb-buffer-page might enable INNODB_BUFFER_PAGE plugin. Or it might enable INNODB_BUFFER_PAGE_LRU plugin - depending on which plugin will loaded (and will see and consume the option) first. And one cannot possibly disable INNODB_BUFFER_PAGE plugin, if the server will try to load INNODB_BUFFER_PAGE_LRU first. I see few solutions for this issue: 1. support a special suffix to option names that will disable prefix matching, for example: mysqld --enable-innodb-buffer-page! this will not match innodb-buffer-page-lru, so it's unambiguous. Of course, any dedicated character can be used, --enable-innodb-buffer-page$, or even a prefix --enable-exact-innodb-buffer-page, --enable-strict-innodb-buffer-page 2. add a special command-line option to disable prefix matching: mysqld --disable-getopt-prefix-matching --enable-innodb-buffer-page 3. do like mysql-5.7 is doing and completely remove support for prefix matching. In 5.7 one has to specified option names in full. I suspect that this is likely to break many exising setups. Opinions? Anybody cares about this obscure issue at all? Regards, Sergei

3 3

[Maria-developers] Buildslave terrier2 disabled due to KVM problems, needs fixing to be re-enabled
by Kristian Nielsen 05 Jan '15

05 Jan '15

Jani, Daniel, I discovered that our buildslave terrier2 seems to have problems with its KVM. Builds are incredibly slow, and it appears that it is not using KVM at all. So it is apparently using just QEMU, and soft-emulating the CPU. This is too slow to be useful, I think. So for now, I've removed all usage of terrier2 from the Buildbot config. If one of you wants to fix terrier2 to be useful again, then the patch to revert (to re-enable terrier2) is this: revid:knielsen@knielsen-hq.org-20150105182519-7492d95v1blss8jf http://bazaar.launchpad.net/~maria-captains/mariadb-tools/trunk/revision/kn… Before re-enabling, KVM needs to be somehow fixed to work (eg. something like `time perl -le 'for (0..100000000){}'` needs to be tested to run at about the same speed inside a guest VM as directly on the host). Maybe it would be best to upgrade the OS on the machine, maybe to the same as we have on bb01/bb02? We have had a number of problems on terrier2 with various KVM issues over the time. - Kristian.

1 0

[Maria-developers] New buildbot? - debpkg101-sid from Otto's github
by Otto Kekäläinen 02 Jan '15

02 Jan '15

Hello Daniel! Thanks for setting up the current buildbots that build my branches of 5.5 and 10.0. Could you now please also set up one the build from my 10.1 branch as I have started working on it too? The buildbot should be a copy of http://buildbot.askmonty.org/buildbot/builders/debpkg-sid/ but so that it pulls in the branch 10.1 from the repository https://github.com/ottok/mariadb Thanks! 2014-09-01 11:25 GMT+03:00 Otto Kekäläinen <otto(a)seravo.fi>: > Hello! > > 2014-03-14 12:59 GMT+02:00 Kristian Nielsen <knielsen(a)knielsen-hq.org>: >> You can see the two builders here: >> >> http://buildbot.askmonty.org/buildbot/builders/debpkg-trusty >> http://buildbot.askmonty.org/buildbot/builders/debpkg-sid/ > > Thanks Kristian and Daniel for creating these in March. > > The trusty build testing is less relevant nowadays while testing 10.0 > has become relevant. > > Could you please delete the buildbot debpkg-trusty and create a copy > of debpkg-sid called debpkg-10.0-sid which would be exactly like > debpkg-sid, but follow commits in this repo: > https://github.com/ottok/mariadb-10.0 ? > > Thanks! > > PS. I already configured the webhook in Github to be called in the > mariadb-10.0 repo as it is done in the mariadb-5.5 repo.

1 2