Re: [Maria-discuss] Expected release of 10.0.28

[Sergei Golubchik]

Hi, Colin!

On Oct 05, Colin Charles wrote:
> While this isn't authoritative, if you go to jira.mariadb.org, you'll see
> it's targeted to be released at 24 November 2016 (but the galera release
> claims 13 Oct 2016). This is how I figure out what's coming out next...

I've fixed the 10.0.28-galera release date, thanks.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Critical Update for CVE-2016-6662

[Sergei Golubchik]

Hi, Alex!

On Sep 13, Alex wrote:
>  From what i noticed , centos6 hosts that were on mysql 5.6 , or mariadb 
> 10.1.17 is using the mysqld_safe.
> Upgraded centos7 hosts , and mysqld_safe is no longer a running process 
> for mariadb 10.1.17.
> 
> Would this mean that only the hosts that do not run the mysqld_safe are 
> safe ?

No, that could be a coincidence.

It is true that the necessary part of the exploit is to run mysqld_safe.
If you use systemd - this particular exploit won't work.

But the vulnerability was fixed in 10.1.17, so even if you'd run
mysqld_safe in 10.1.17 - you would've been safe.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Critical Update for CVE-2016-6662

[Sergei Golubchik]

Hi, Alex!

On Sep 12, Alex wrote:
> Hello,
> 
> In regards to this zero day remote exploit , it seems MariaDB is also 
> affected. Percona seems to have released new versions out to fix this.
> Any news from MariaDB side ?
> 
> http://legalhackers.com/advisories/MySQL-Exploit-Remote-Root-Code-Execution-Privesc-CVE-2016-6662.html

Yes, it was https://jira.mariadb.org/browse/MDEV-10465,
fixed in 5.5.51, 10.0.27, 10.1.17, all released last month.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] howto get MariaDB T-Shirts and Hoodies?

[Sergei Golubchik]

Hi, Lukas!

On Aug 21, Lukas Lehner wrote:
> Hi
> 
> I have lots (> 10) of MongoDB T-Shirts, but I prefer to run MariaDB. Howto
> get cool MariaDB Shirts and Hoodies?
> 
> Witch a search engine I found only "Support MariaDB" products (CafePress).
> That's not my taste :-(

It seems like CafePress is the only way now, short of attending a
conference. But may be we can put more t-shirts there, something you
might like...

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Backporting bug fixes to 1.0?

[Sergei Golubchik]

Hi, Justin!

On Aug 24, Justin Swanhart wrote:
> I have been fired for speaking out about the GPL and MariaDB actions
> that have caused great harm to our ecosystem.

I'm sorry to hear that. Where were you fired from?

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] maxscale 2.0 bug fixes

[Sergei Golubchik]

Hi, Federico!

On Aug 17, Federico Razzoli wrote:
> 2) At some point MaxScale 2.0 will be open and 3.0 will be released
> but not open. When a software exists in both an open and a closed
> source, I expect the open form to be low-quality. This applies to all
> examples I can think of. I will feel a bit better if you could explain
> why you will maintain 2.0 for a reasonable number of years.

I think the MariaDB Corporation would continue to maintain 2.0 for the
very same reason many other companies consider Open Source a good
business strategy - this is good for adoption. And it adds credibility
both to MariaDB Corporation and BSL in particular - which are valid
reasons too.

While I don't know what MariaDB Corporation will decide to do in two
years, I do believe there are direct business benefits in supporting
MariaDB MaxScale 2.0 for a reasonable number of years after it becomes
GPL.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Help on ERROR 1819 (HY000): Your password does not satisfy the current policy requirements

[Sergei Golubchik]

Hi, Karthick!

On Jul 03, Karthick Subramanian wrote:
> Dear Experts,
> 
> I almost tried all sort of combinations that I think of very difficult to
> crack, but repeatedly failed to satisfy MariaDB not throw this error.
> 
> Can anyone guide me on what is the specific needs for successful password.
> 
> Warning message:
> 
> +-+--++
> | Level   | Code | Message
> |
> +-+--++
> | Warning | 1819 | cracklib: error loading dictionary 
> |
> | Error   | 1819 | Your password does not satisfy the current policy 
> requirements |
> +-+--++

Well, it says that in attempt to validate the password, cracklib has
failed with an error message "error loading dictionary".

May be the dictionary file is missing. May be it is corrupted.
Use SHOW VARIABLES LIKE 'cracklib_password_check_dictionary' to see the
location of the cracklib dictionary.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Orphaned TokuDB tables

[Sergei Golubchik]

Hi, Rich!

Shall we include it in packages together with ha_tokudb.so?
Or is it too dangerous for that?

On Aug 02, Rich Prohaska wrote:
> Hello All,
> 
> Sometimes, TokuDB gets left in a state where it thinks that parts of a
> table exist (as observed in TokuDB's file map), but not all of the
> underlying files (the TokuDB data files and/or the frm file) exist.  This
> probably occurs due to the interaction between the non-transactional DDL
> that MySQL/MariaDB uses to create or drop tables, and the transactional
> TokuDB engine underneath.  There is some bug there.
> 
> I added the tokudb_delete_dnames_from_directory tool to the
> https://github.com/prohaska7/mariadb-server/tree/fix-tokudb-filemap
> branch.  This standalone tool can be used to delete dnames from the
> tokudb.directory.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB Knowledge base - 500 error

[Sergei Golubchik]

Hi, Karthick!

On Jul 27, Karthick Subramanian wrote:
> Did anyone facing this 500 error while accessing the Maria DB Knowledge
> Base. Thank you.

I've just tried it - seems to work for me. And yesterday it worked too.
Did you get this error for any KB page or for some specific url?

> Don't know to whom we need to report this, so sending to this mail ID.
> Sorry if it cause any inconvenient for others.

yeah, silly enough, the contact information is in the kb:
https://mariadb.com/kb/en/meta/contact/
"
  If you have a question or comment about this site, please join the
  #maria channel on freenode or email us, kb-admins [AT] mariadb (dot)
  org.
"
> When we click the Support in MariaDb website, getting below error:
> 
> Database Error:
> 
> *There seems to be a problem connecting to the database server
> specified in your configuration file. Please contact your local system
> administrator for further assistance. *

Okay, support portal is not part of the KB. But anyway, I've tried
clicking Suport and it worked for me too.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Finicky syntax with RETURNING?

[Sergei Golubchik]

Hi, Rhys.Campbell!

On Jul 27, rhys.campb...@swisscom.com wrote:
> Hi Sergei,
> 
> Thanks for your response. Shouldn't this work?
> 
> EXPLAIN DELETE uc FROM t1 AS uc
> WHERE uc.id IN (SELECT p.id FROM t2 AS p WHERE p.account_id = )
> RETURNING uc.some_id;
> 
> Or does that count in the definition of a multi-table delete? (I thought 
> not). 

It does. See the manual: https://mariadb.com/kb/en/mariadb/delete/

Single table delete: DELETE FROM ...
Multi-table detele: DELETE tables FROM

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Creating Custom Storage Engine

[Sergei Golubchik]

Hi, Jonathan!

On Jul 20, Jonathan Ellithorpe wrote:
> 
> Hmm... thanks for pointing that out about system metadata caching. I'll
> have the solve the problem of MariaDB thinking something exists
> (table/index/...) when it doesn't, or thinking that something doesn't exist
> when it does

We have a solution for that, see
https://mariadb.com/kb/en/mariadb/table-discovery/

It was initially created exactly for this problem - when many MySQL
servers connect to the same NDB cluster, and the table is created on one
node and used on another.

Note that to support for SQL semantics you might need some kind of a
distributed lock manager. For example, the server does not expect table
metadata to change *while the UPDATE is running*. Or if SELECT needs to
do multiple table scans, it expects to see the same set or rows in all
these table scans.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Creating Custom Storage Engine

[Sergei Golubchik]

Hi, Jonathan!

On Jul 18, Jonathan Ellithorpe wrote:
> Hi All,
> 
> I searched around but couldn't find documentation / tutorials on creating a
> custom storage backend for MariaDB. Could anyone point me in the right
> direction?

Storage engine examples in the source tree.
"MySQL 5.1 Plugin Development" book.
https://mariadb.com/kb/en/mariadb/storage-engines-storage-engine-development/

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Key lengths for file_key_management plugin

[Sergei Golubchik]

Hi, Honza!

On Jul 19, Honza Horak wrote:
> On 07/15/2016 11:31 AM, Sergei Golubchik wrote:
> >
> > On Jul 14, Honza Horak wrote:
> >> I've asked directly in the documentation page, but maybe someone will
> >> know answer here:
> >> https://mariadb.com/kb/en/mariadb/data-at-rest-encryption/#comment_2346
> >>
> >> In that article it is said that 128, 192 or 256-bit keys are
> >> supported, but later AES_CTR and AES_CBC modes talk only about
> >> 128bit keys. What piece of information am I missing?
> >
> > The article says "the plugin will use AES with the 128-bit keys in
> > the CTR mode for encrypting tablespace pages".
> >
> > Because tablespace pages are always encrypted with a 128-bit
> > tablespace key. Which is different for every tablespace and it is
> > generated from the user-specified key, which might be 128-, 192-, or
> > 256-bit.
> >
> > Temporary files and binary logs are encrypted directly with the
> > user-specified key of a user-specified length.
> 
> Thanks for the answer, but one more question to make sure I understand
> it correctly -- in case of choosing CBC mode, 128 keys are used only
> for tablespace pages or also for temporary files and binary logs?

Depends on your keys that you've specified in a file. Say, if you've put
192-bit keys there, then 192-bit keys will be used to encrypt temporary
files and binary logs. And 192-bit key will be used to generate a
128-bit tablespace key, which in turn will used to encrypt a tablespace.
It does not depend on the mode, CBC or CTR.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] mariadb-connect-engine-10.1 : Depends: unixODBC but it is not installable

[Sergei Golubchik]

Hi, Umarzuki!

On Jul 13, Umarzuki Mochlis wrote:
> 2 of 3 mariadb servers 10.1, ubuntu 16.04 unable to install
> mariadb-connect-engine-10.1 with error
> 
> The following packages have unmet dependencies:
>  mariadb-connect-engine-10.1 : Depends: unixODBC but it is not installable
> E: Unable to correct problems, you have held broken packages.
> 
> I fould bugreport at https://jira.mariadb.org/browse/MDEV-10032
> 
> but it seems stalled?

It's closed, the jira page above says that it was fixed in 10.1.15.

But we pulled that release off, it's not in repos.

You'll get a fix for this bug in 10.1.16, which should be released early
next week.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Key lengths for file_key_management plugin

[Sergei Golubchik]

Hi, Honza!

On Jul 14, Honza Horak wrote:
> I've asked directly in the documentation page, but maybe someone will 
> know answer here:
> https://mariadb.com/kb/en/mariadb/data-at-rest-encryption/#comment_2346
> 
> In that article it is said that 128, 192 or 256-bit keys are supported, 
> but later AES_CTR and AES_CBC modes talk only about 128bit keys. What 
> piece of information am I missing?

The article says "the plugin will use AES with the 128-bit keys in
the CTR mode for encrypting tablespace pages".

Because tablespace pages are always encrypted with a 128-bit tablespace
key. Which is different for every tablespace and it is generated from
the user-specified key, which might be 128-, 192-, or 256-bit.

Temporary files and binary logs are encrypted directly with the
user-specified key of a user-specified length.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Include TokuDB Hot Backup in the bundled plugin list

[Sergei Golubchik]

Hi, Jocelyn!

Yes, it's MDEV-8843.
The plan is to do it in 10.1, but it's unlikely to happen in time for
the 10.1.15.

On Jun 25, jocelyn fournier wrote:
> Hi,
> 
> 
> In MariaDB, TokuDB is bundled as a plugin, but it's not the case for the 
> Hot Backup plugin (https://github.com/percona/tokudb-backup-plugin).
> 
> I have to compile it manually against the server source code to have it 
> working properly.
> 
> Would it be possible to include it in the bundled plugins ?
> 
> Thanks!
>Jocelyn
> 
Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Event types in performance schema

[Sergei Golubchik]

Hi, Rhys.Campbell!

On Jun 07, rhys.campb...@swisscom.com wrote:
> 
> I'm doing some stuff with the performance_Schema and have a quick
> question about event types.
> 
> These appear in the events_statements_summary_by_account_by_event_name
> table (and others) and look like...
> 
> statement/com/Ping
> statement/sql/select
> statement/sql/insert
> statement/sql/commit
> statement/sql/update
> 
> Am I correct in thinking the COMMIT type would be a total of all the
> events of a transaction.

No, these are _sql statements_. So, statement/sql/commit event happens
when you issue an SQL statement "COMMIT", it does not include any other
SQL statements.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] ALGORITHM INPLACE for 10.0.24-MariaDB

[Sergei Golubchik]

Hi, Rhys.Campbell!

On May 04, rhys.campb...@swisscom.com wrote:
> Hi All,
> 
> For the following statement...
> 
> ALTER ONLINE TABLE dom_audit_event ENGINE=INNODB, ALGORITHM=INPLACE;
> 
> I can still see that MariaDB is still using the value of tmpdir. We
> have this set to tmpfs on some of our servers and this causes some
> ALTER statements on big tables to fail. It was my understanding that
> if you specified INPLACE then the statement would fail if this was not
> possible.

Yes, your understanding is correct.

Furthermore, copy (not inplace) ALTER does not use tmpdir, it creates a
temporary table in the datadir, and later renames it to the actual table
name. If the temporary table were created in tmpdir, it could not be
renamed to a different filesystem.

So, that tmpdir usage you're seeing is not due to not-inplace alter.
The alter isstill done inplace, and tmpdir is used for something else.
Probaby for merge sort.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Dynamically rewrite query

[Sergei Golubchik]

Hi, Erik!

On Apr 20, Erik Cederstrand wrote:
> 
> CREATE TABLE `t1` ( `i` int(11) NOT NULL, `j` int(11) NOT NULL, `l`
> int(11) NOT NULL, `m` int(11) DEFAULT NULL, PRIMARY KEY
> (`i`,`j`,`l`));
> 
> CREATE TABLE `t2` ( `i` int(11) NOT NULL, `k` int(11) DEFAULT NULL,
> PRIMARY KEY (`i`)) ENGINE=InnoDB DEFAULT CHARSET=utf8;
> 
> The problematic query is this:
>SELECT DISTINCT t2.i, t2.k, t1.j, t1.l, t1.m FROM t2 LEFT JOIN t1
>ON t2.i=t1.i WHERE t2.k < 123;
> 
> The query is slow because the DISTINCT clause forces the query to use
> a temporary table. Removing DISTINCT makes the query 100x faster.
> ... 
> it could also be argued that MariaDB should be smart enough to detect
> that the DISTINCT is unnecessary and ignore it (should I create a bug
> report for that?).

Yes, please, do report it! On the first glance this looks like a rather
simple optimization.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] expire_logs_days

[Sergei Golubchik]

Hi, Reindl!

On Apr 05, Reindl Harald wrote:
> i find a config variable with "days" just annoying - why - because in
> case of thin-provisioned virtual servers it wastes space in
> environment where the slave is for sure not down longer then a few
> hours and if it is longer down it's just my problem to act
> 
> http://dev.mysql.com/doc/refman/5.7/en/server-system-variables.html#sysvar_expire_logs_days

I think we can change it from int to float, and you'll be able to
specify values, like, 0.1, if you'd like. I believe it's a backward
compatible change, so it should not break anything.

Care to create a feature request at jira.mariadb.org?

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] The insert performance issue

[Sergei Golubchik]

Hi, 林澤宇!

On Mar 10, 林澤宇 wrote:
> Hi ALL ,
> 
> I test the MariaDB performance of insert command .
> I use a file include 10 insert command statements to test the insert
> performanc .
> On local server ,the Mariadb spent about 20 second to insert data;but on
> remote server ,the MariaDB spent about 30 second to insert data .
> I execute the command " mysql -h db1 -utest -ptest -D IST <
> insert_file.sql" on local and remote servers .
> 
> Why the MariaDB has about 10 second gap ?
> Maybe the network should to cause some latancy ,but the time should not
> have so long .
> 
> What issue to cause the gap ?

May be hostname resolution? As far as I understand the default resolver
timeout is 5 seconds (man resolv.conf) - two failed name lookups gives
you exactly 10 seconds.

Try --skip-name-resolve on the remote mariadb server and make sure you
only use IP adresses when granting privileges and when connecting to the
server.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] How to rollback XA Transactions

[Sergei Golubchik]

Hi, Marc!

While not exactly user-friendly, you can still run

$ mysql -e 'xa recover'|hexdump -C

You'll see hex data there.

Regards,
Sergei

On Feb 22, Marc Schöchlin wrote:
> Hi Justin,
> 
> how can i get the hex representation of the the transaction identification 
> strings?
> ("XA RECOVER CONVERT XID" seems only to be available on MySQL 5.7)
> 
> >
> > I have not tested this, but I think you will need to convert the strings to 
> > hex then do:
> > xa rollback 0xabcdef <-- where 0xabcdef is the hex representation of the 
> > string.
> >
> > --Justin
> >
> > On Sat, Feb 20, 2016 at 5:53 AM, Marc Schöchlin  > > wrote:
> >
> > Hi,
> >
> > i currently have the situation that i cannot modify tables/indices beaus of 
> > XA transactions which are in progress.
> >
> > mysql://root@localhost:/(none) - 11:37:55 > xa recover;
> > +--+--+--+--+
> > | formatID | gtrid_length | bqual_length | data 
> > |
> > +--+--+--+--+
> > |  4871251 |   35 |   29 | ^W<89>^B 
> > XF+foo-bar-de05,server,P4837foo-bar-de05,server,P4837,^A |
> > |  4871251 |   35 |   29 | ^N<89>^B 
> > XF+foo-bar-de05,server,P4837foo-bar-de05,server,P4837,^A |
> > +--+--+--+--+
> > 2 rows in set (0.00 sec)
> >
> > The documentation (https://mariadb.com/kb/en/mariadb/xa-transactions/) 
> > suggests that xa transaction can be rolled back by using "xa rollback xid" 
> > but it seems that this is not possible because of the non-ascii characters.
> > As you can see in the attached Innodb status, these transactions a active 
> > for a log time now and should not be relevant anymore.
> > Naturally restarting the server should not help.

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] kb login doesn't work!

[Sergei Golubchik]

Hi, Justin!

On Feb 22, Justin Swanhart wrote:
> I said it returns a 500!
> 
> I don't know why it doesn't work for my account.
> 
> And it doesn't change the fact that you can't reply to messages from
> this mailing list.  How hard is it to make it do so? Seriously, others
> have complained about this system (have to use web for something we
> get in mail) too and it is not working for us.  This web/mail
> Frankenstein monster has not been working for us for a very long time.
> FIX IT.  PRETTY PRETTY PLEASE!

I've passed it over to the web team.
There was a bug, it was fixed.
Thanks for reporting it! You account should be fixed now too.

As for the accepting kb replies by email - it will take longer than a
bug fix, but we plan to look into that.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Adding timestamp column to a system

[Sergei Golubchik]

Hi, Peter!

On Jan 29, Peter Mclarty wrote:
> Hi All
> Running Mariadb 10.0.16 and need to add a timestamp column to a table with
> has around 60m rows
> 
> Testing on lesser hardware instances suggest around 40 minutes to make this
> change on the production system.
> 
> To reduce the down time I was planning to alter the slave and role swap the
> slave to become the master
> 
> I have added the column to a test slave and notice it alway populates as
> Unix_epoch 0 ie no current timestamp
> date   |  timestamp
> 2016-01-29 | -00-00 00:00:00
> 
> This sounds about right, but to confirm it's not a bug.

Depends on what you want. If you do

  ALTER TABLE ... ADD COLUMN x TIMESTAMP DEFALT CURRENT_TIMESTAMP

you'll get current  timestamp;

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] mariadb on freebsd: spinlocks or mutex?

[Sergei Golubchik]

Hi, Federico!

On Jan 21, Federico Razzoli wrote:
> While compiling MariaDB on FreeBSD, I've noted that it can optionally
> be compiled with spinlocks instead of mutexes. What's the best choice?
> Any experience?

Yes, this option enables MariaDB "fast mutexes".

They're disabled by default for years and completely removed in MySQL
(5.7 or 5.6? - don't remember) and in MariaDB 10.2.

Don't use them.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Select optimizations

[Sergei Golubchik]

Hi, Roberto!

On Jan 20, Roberto Spadim wrote:
> Hi guys! There's a "easy/fast" way to rewrite a query from this:
> 
> Select datetime,key,value from table where ...
> 
> To this:
> 
> Select datetime, value1 as key1, value2 as key2,  From table where ...
> 
> I don't remember the right name in R/python (pandas) language but it's
> something like vector/matrix merge
> 
> The ideia is, instead of one line of date key vale, use a single date, and
> "put" values as columns, and the columns name is the key name
> 
> The table structure is something like:
> Create table (
> Datetime datetime not null,
> Key varchar(255) not null,
> Value double not null,
> Primary key(datetime,key)
> )
> 
> Any idea is wellcome
> I know there's many solutions, if i have >100 columns, if datetime is
> sparse (only one key have values per datetime value), etc etc

Check the PIVOT table type of the CONNECT engine.
https://mariadb.com/kb/en/mariadb/connect-table-types-pivot-table-type/

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org
-- 
Vote for my Percona Live 2016 talks:
https://www.percona.com/live/data-performance-conference-2016/sessions/mariadb-connectors-fast-and-smart-new-protocol-optimizations#community-voting
https://www.percona.com/live/data-performance-conference-2016/sessions/mariadb-101-security-validation-authentication-encryption#community-voting

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] mariadb still uses indexes with high average value group size.

[Sergei Golubchik]

Hi, dE!

On Jan 02, dE wrote:
> Hi!
>  I was playing around with indexes after we faced problems with 
> mariadb choosing the wrong index because of high average value group size.
> 
> So I created a table with 4 text columns and indexed the 4th column with 
> a prefix length of 1.
> 
> Following are the stats in brief --
> 
> The index has 35 unique values
> There are around 104 rows in total.
> this will give an average value group size of 29714
> Cardinality is calculated as / of value groups>, thus it's value must be 35
> 
> BUT from show indexes, it's value is shown as 72 
> (innodb_stats_persistent_sample_pages = 999).

It's because InnoDB lies about the cardinality. It reports twice the
actual value to the server. If MyISAM would report the cardinality as
36, InnoDB will say it's 72.

You can enable the Engine Independent Table Statistics in MariaDB:
https://mariadb.com/kb/en/mariadb/engine-independent-table-statistics/
This will show the correct value.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org
-- 
Vote for my Percona Live 2016 talks:
https://www.percona.com/live/data-performance-conference-2016/sessions/mariadb-connectors-fast-and-smart-new-protocol-optimizations#community-voting
https://www.percona.com/live/data-performance-conference-2016/sessions/mariadb-101-security-validation-authentication-encryption#community-voting

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.23 now available

[Sergei Golubchik]

Hi, Reindl!

On Dec 19, Reindl Harald wrote:
> 
> [root@buildserver:~]$ mysql-test.sh
> Logging: ./mysql-test-run.pl  --default-myisam --parallel=1 
> --max-test-fail=0 --mysqld=--binlog-format=mixed --skip-ndb --force
> Invalid option "--default-myisam"
> 
> seriously in a minor-update?

You're right. It might've been better to do it in 10.2.

On the other hand, it's an option that never did anything, it was always
ignored and shouldn't have been merged in the first place. And it's not
like mysql-test-run is used in production, it's purely for testing.

But still, I agree, better to do changes like this before GA.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org
-- 
Vote for my Percona Live 2016 talks:
https://www.percona.com/live/data-performance-conference-2016/sessions/mariadb-connectors-fast-and-smart-new-protocol-optimizations#community-voting
https://www.percona.com/live/data-performance-conference-2016/sessions/mariadb-101-security-validation-authentication-encryption#community-voting

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.1.8 Stable (GA) now available

[Sergei Golubchik]

Hi, Honza!

On Dec 09, Honza Horak wrote:
> Just a friendly poke to see whether there are some news about 10.1 
> support/EOL plans..

Thanks and sorry for not getting to you earlier.

We've just decided to support all GA releases for five years, so 10.1
will EOL at the end of 2020.

But the current plan is to do regular releases only for the latest three
GA releases (at the moment it's 5.5, 10.0, and 10.1) and the other two
will be released on demand, when there's a security issue or a critical
bug to fix.

Does that work for you?

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org
-- 
Vote for my Percona Live 2016 talks:
https://www.percona.com/live/data-performance-conference-2016/sessions/mariadb-connectors-fast-and-smart-new-protocol-optimizations#community-voting
https://www.percona.com/live/data-performance-conference-2016/sessions/mariadb-101-security-validation-authentication-encryption#community-voting

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] mysql_install_db as user

[Sergei Golubchik]

Hi, Felipe!

On Nov 12, Felipe Gasper wrote:
> Hello,
> 
> Is this script supposed to be runnable as a normal (i.e., non-root)
> user?
> 
> In MySQL 5.6 there is logic that prevents the chown() of the mysql
> data dir when running as non-root. MariaDB 10.1.8 doesn’t seem to have
> this logic, so the script fails when chown() fails.
> 
> Is this change intended? We have tests that verify functionality of
> user-created, UNIX-socket-only MySQL/MariaDB servers. These work on
> MySQL 5.6 but not on MariaDB 10.1.

There's no special logic around it in MySQL 5.6 and no changes in
MariaDB 10.1.

chown is done if user=xxx is specified on the command line or in the
my.cnf file. And if chown fails, you'll see the error message and
mysql_install_db will abort. Both in MySQL and MariaDB.

What is different, though, in MySQL mysql_install_db will only read
the [mysqld] section in my.cnf, while in MariaDB it will read all
sections that the server will (for me it's: mysqld server mysqld-10.1
mariadb mariadb-10.1 client-server galera). So if one of those sections
in your my.cnf contains "user=mysql", then MariaDB's mysql_install_db
will see it, but MySQL's mysql_install_db won't.

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] mysql_install_db as user

[Sergei Golubchik]

Hi, Felipe!

On Nov 13, Felipe Gasper wrote:
> >
> > There's no special logic around it in MySQL 5.6 and no changes in
> > MariaDB 10.1.
> >
> > chown is done if user=xxx is specified on the command line or in the
> > my.cnf file. And if chown fails, you'll see the error message and
> > mysql_install_db will abort. Both in MySQL and MariaDB.
> 
> Hi Sergei! Thanks for responding. :)
> 
> I wonder, are we looking at the same things?

Ah, right. We are not. I was comparing shell script to a shell script.
And perl script to a perl script. You've compared MySQL-5.6 perl script
to MariaDB 10.1 shell script :)

> The 5.6 box’s mysql_install_db has:
> -
>if ($opt_user and -w "/")
>{
>  chown($pwnam[2], $pwnam[3], $dir)
>or error($opt, "Could not chown directory $dir");
>}
> -

In MariaDB 10.1 mysql_install_db.pl has

-
  chown($opt->{user}, $dir) if -w "/" and !$opt->{user};
-

> … whereas the MariaDB box has:
> -
>  chown $user "$dir"
>  if test $? -ne 0
>  then
>echo "Cannot change ownership of the database directories to the 
> '$user'"
>echo "user.  Check that you have the necessary permissions and try 
> again."
>exit 1
>  fi
> --

In MySQL 5.6 mysql_install_db.sh has

-
chown $user $dir
if test $? -ne 0
then
  echo "Cannot change ownership of the database directories to the '$user'"
  echo "user.  Check that you have the necessary permissions and try again."
  exit 1
fi
-

which is almost exactly the same (only differece - it'll fail if $dir
contains spaces)

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] TokuDB + virtual column

[Sergei Golubchik]

Hi, Roberto!

On Nov 03, Roberto Spadim wrote:
> Hi guys, there's a plan to include virtual column to tokudb storage?
> 
> 
> /* Erro SQL (1910): TokuDB storage engine does not support computed columns
> */

It's MDEV-6186.

As for the _plan_ - I don't know yet, ask again in a couple of weeks :)

Regards,
Sergei
Chief Architect MariaDB
and secur...@mariadb.org

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.22 now available

[Sergei Golubchik]

Hi, Reindl!

Could you show more of the build log please? You've cut the compiler
error out.

On Oct 30, Reindl Harald wrote:
> seems to be broken - at least i don't expect anything to for a minor 
> release then just chnage the version line in a RPM-SPEC and fire 
> "rpmbuild -bb"
> 
> [ 45%] Building CXX object 
> storage/xtradb/CMakeFiles/xtradb.dir/log/log0log.cc.o
> cd /home/builduser/rpmbuild/BUILD/mariadb-10.0.22/storage/xtradb && 
> /usr/bin/c++   -DHAVE_CONFIG_H -DHAVE_IB_ATOMIC_PTHREAD_T_GCC=1 
> -DHAVE_IB_GCC_ATOMIC_BUILTINS=1 -DHAVE_IB_GCC_ATOMIC_BUILTINS_64=1 
> -DHAVE_IB_GCC_ATOMIC_BUILTINS_BYTE=1 -DHAVE_IB_GCC_ATOMIC_THREAD_FENCE=1 
> -DHAVE_IB_GCC_SYNC_SYNCHRONISE=1 -DLINUX_NATIVE_AIO=1 
> -DSIZEOF_PTHREAD_T=8 -m64 -O2 -g0 -march=sandybridge -mtune=sandybridge 
> -fopenmp -mfpmath=sse -pipe -fno-strict-aliasing -fomit-frame-pointer 
> -finline-functions -fexceptions -fstack-protector-strong 
> --param=ssp-buffer-size=6 -D_FORTIFY_SOURCE=2 -Wstack-protector -Wformat 
> -Werror=format-security -O3 -funroll-loops -funswitch-loops 
> -minline-all-stringops -fwrapv -fno-stack-protector -fstack-protector 
> --param=ssp-buffer-size=8 -fPIC -fno-exceptions -D_GNU_SOURCE 
> -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE  -fPIC -fno-exceptions 
> -fno-rtti -O3 -g -static-libgcc -fno-omit-frame-pointer 
> -fno-strict-aliasing -Wno-uninitialized -DDBUG_OFF 
> -DMY_PTHREAD_FASTMUTEX=1 
> -I/home/builduser/rpmbuild/BUILD/mariadb-10.0.22/include 
> -I/home/builduser/rpmbuild/BUILD/mariadb-10.0.22/storage/xtradb/include 
> -I/home/builduser/rpmbuild/BUILD/mariadb-10.0.22/storage/xtradb/handler 
> -I/home/builduser/rpmbuild/BUILD/mariadb-10.0.22/sql-DUNIV_LINUX 
> -D_GNU_SOURCE=1  -fvisibility=hidden -o 
> CMakeFiles/xtradb.dir/log/log0log.cc.o -c 
> /home/builduser/rpmbuild/BUILD/mariadb-10.0.22/storage/xtradb/log/log0log.cc
> storage/xtradb/CMakeFiles/xtradb.dir/build.make:1142: recipe for target 
> 'storage/xtradb/CMakeFiles/xtradb.dir/handler/xtradb_i_s.cc.o' failed

This is what make says. Compiler error was about xtradb_i_s.cc and it
was earlier in the log.

> make[2]: *** [storage/xtradb/CMakeFiles/xtradb.dir/handler/xtradb_i_s.cc.o] 
> Error 1
> make[2]: *** Waiting for unfinished jobs

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.22 now available

[Sergei Golubchik]

Hi, Reindl!

On Oct 30, Reindl Harald wrote:
> Am 30.10.2015 um 09:20 schrieb Sergei Golubchik:
> >
> > Could you show more of the build log please? You've cut the compiler
> > error out.
> 
> output of "rpmbuild -bb mariadb.spec >> mariadb.log 2>> mariadb.log"
> https://access.thelounge.net/harry/mariadb.log.tar.bz2

Okay, found it, thanks.

I'm pushing the fix now. But as a workaround you can remove
-DENABLED_PROFILING=OFF from your specs file.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Error when migrating from TokuDB 7.5.2 to MariaDB 5.5.44 (works with 5.5.41)

[Sergei Golubchik]

Hi, Fernando!

On Sep 28, Fernando Marcelo wrote:
> I'm trying to migrate a TokuDB 7.5.2 server to MariaDB 5.5.44.
> 
> Since I can't stop the production server, I got a snapshot of the running
> instance. So MariaDB will have to recover.
> 
> Unfortunately i get an error when TokuDB recovery is running in version
> 5.5.44.
> 
> Previously I tried with MariaDB 5.5.41 which worked fine. But since I'm
> upgrading, I would like to use the latest version.
> 
> Mon Sep 28 03:53:37 2015 TokuFT recovery bw_begin_checkpoint at 158147092471 
> timestamp 1443133470008593 (bw_between)
> Mon Sep 28 03:53:37 2015 TokuFT recovery turning around at begin checkpoint 
> 158147092471 time 181293960
> Mon Sep 28 03:53:37 2015 TokuFT recovery starts scanning forward to 
> 158147993496 from 158147092471 left 901025 (fw_between)
> /home/buildbot/buildbot/build/mariadb-5.5.44/storage/tokudb/ft-index/ft/serialize/ft-serialize.cc:275
>  deserialize_ft_versioned: Assertion `readsz >= translation_size_on_disk' 
> failed (errno=2)

This is an internal tokudb assertion.
I'd say it's *very* unlikely that anything we've done in MariaDB
could've caused this assertion.

But MariaDB-5.5.41 included tokudb-7.5.3, and MariaDB-5.5.44 included
tokudb-7.5.7. It's quite possible that some tokudb change between 7.5.3
and 7.5.7 is the reason.

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] what is adding the -g compile parm?

[Sergei Golubchik]

Hi, Reindl!

On Sep 20, Reindl Harald wrote:
> Am 20.09.2015 um 19:14 schrieb Sergei Golubchik:
> >
> > On Sep 20, Reindl Harald wrote:
> >> "-O3 -g" appears in the cmake output, that likely breaks ld.gold / -flto
> >> and so the question: who the explain the build *not* add that useless
> >> param for production builds like most oterh software don't
> >>
> >> without one could also remove the strip-calls in rpm-spec-files!
> >
> > Sorry, I have more questions that answers...
> >
> > * How do you build?
> 
> + export 'CFLAGS=-m64 -O2 -g0 -march=sandybridge -mtune=sandybridge 
> -fopenmp -mfpmath=sse -pipe -fno-strict-aliasing -fomit-frame-pointer 
> -finline-functions -finline-limit=60 -fexceptions 
> -fstack-protector-strong --param=ssp-buffer-size=6 -D_FORTIFY_SOURCE=2 
> -Wstack-protector -Wformat -Werror=format-security -O3 -funroll-loops 
> -funswitch-loops -minline-all-stringops -fwrapv -fno-stack-protector 
> -fstack-protector --param=ssp-buffer-size=8 -fPIC -fno-exceptions 
> -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE'

But you cannot be building this this line, right?
You invoke cmake, somehow, and make, somehow. How?
And may be you set CXXFLAGS and I don't know what else.

> > Where do you see -O3 -g?
> 
> the " -fPIC -O3 -g -static-libgcc" stuff comes *after* my global CFLAGS 
> and so overrides them

Okay, that's better.
"static-libgcc" is only used in two files in the source tree:
./BUILD/compile-bintar and ./cmake/build_configurations/mysql_release.cmake

So, either you're using BUILD/compile-bintar script (which I find
unlikely) or you're invoking cmake with -DBUILD_CONFIG=mysql_release
(which I find likely - and that's exactly why I was asking "how do you
build" !)

This is not configurable, you cannot switch a part of
mysql_release.cmake off. Either you use -DBUILD_CONFIG=mysql_release or
you don't. If you don't like it - you can simply add all options from it
manually to your CFLAGS and then you won't need -DBUILD_CONFIG

But you said it conflicts with -flto. Where does -flto come from? I
don't see it in your CFLAGS. Nor is it on the compiler's command line
below:

> [  6%] Building C object strings/CMakeFiles/strings.dir/ctype-cp932.c.o
> cd /home/builduser/rpmbuild/BUILD/mariadb-10.0.21/strings && /usr/bin/cc 
>   -DDISABLE_MYSQL_THREAD_H -DHAVE_CONFIG_H -m64 -O2 -g0 
> -march=sandybridge -mtune=sandybridge -fopenmp -mfpmath=sse -pipe 
> -fno-strict-aliasing -fomit-frame-pointer -finline-functions 
> -finline-limit=60 -fexceptions -fstack-protector-strong 
> --param=ssp-buffer-size=6 -D_FORTIFY_SOURCE=2 -Wstack-protector -Wformat 
> -Werror=format-security -O3 -funroll-loops -funswitch-loops 
> -minline-all-stringops -fwrapv -fno-stack-protector -fstack-protector 
> --param=ssp-buffer-size=8 -fPIC -fno-exceptions -D_GNU_SOURCE 
> -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE  -fPIC -O3 -g -static-libgcc 
> -fno-omit-frame-pointer -fno-strict-aliasing  -Wno-uninitialized 
> -DDBUG_OFF -DMY_PTHREAD_FASTMUTEX=1 
> -I/home/builduser/rpmbuild/BUILD/mariadb-10.0.21/include-o 
> CMakeFiles/strings.dir/ctype-cp932.c.o   -c 
> /home/builduser/rpmbuild/BUILD/mariadb-10.0.21/strings/ctype-cp932.c

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] what is adding the -g compile parm?

[Sergei Golubchik]

Hi, Reindl!

On Sep 20, Reindl Harald wrote:
> "-O3 -g" appears in the cmake output, that likely breaks ld.gold / -flto 
> and so the question: who the explain the build *not* add that useless 
> param for production builds like most oterh software don't
> 
> without one could also remove the strip-calls in rpm-spec-files!

Sorry, I have more questions that answers...

* How do you build? Where do you see -O3 -g?
* Why do you think it breaks gold? It shouldn't. In fact, gold doesn't
  work for a completely different reason, see MDEV-5982.
* Why would it break -flto? TokuDB is always compiled with -flto, so
  either "-O3 -g" isn't added autumatically in our builds or -flto works
  with it just fine.

Besides, we don't use rpm spec files, cpack generates then
internally as needed.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] writing plugin with .net language (C#)

[Sergei Golubchik]

Hi, Roberto!

On Sep 18, Roberto Spadim wrote:
> guys , i'm still with doubts... but not more about c#, now about c++
> i was thinking about rewrite the code with c++
> 
> my doubt is could i use C++ 11 ? or i will have problem with mariadb
> makefiles etc etc?

Sure, you  can use any language you want, C++11 is perfectly fine too.
cmake doesn't care.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Cross compile issue

[Sergei Golubchik]

Hi, Johnson!

Thanks a lot!

Based on this thread I've created a page in the knowledge base with the
instructions: https://mariadb.com/kb/en/mariadb/cross-compiling-mariadb/

Hopefully, the next developer who will try cross-compiling MariaDB will
find it almost as simple as normal compilation :)

Regards,
Sergei

On Sep 07, Johnson Cheng wrote:
> Dear All,
> 
> SOLVED.
> I share my final solution here, and hope it can help more people in the 
> future.
...

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] writing plugin with .net language (C#)

[Sergei Golubchik]

Hi, Roberto!

On Sep 13, Roberto Spadim wrote:
> Hi guys,
> I have a program wrote in c#, it process some code and the result is a
> struct (an array of string/double/int)
> I want to include this program inside mariadb server as a daemon plugin (to
> process data) + a storage engine (to read struct data)
> 
> is that possible? considering that i can write dll with .net?
> what about write this to linux (.so files) ?

I think it's possible, why not? You might need a thin layer of the
interface code in C (seeing your code, I believe you can do that easily :),
and the main logic in C#. On Linux you can, perhaps, use Mono.

But I haven't heard of anyone doing that, so I don't know of any
specific pitfalls that you might face on this way.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MySQL 5.7 JSON

[Sergei Golubchik]

Hi, Tom!

On Sep 03, Tom Worster wrote:
> Are MySQL 5.7 JSON features
>  -json>  likely to end up in MariaDB anytime? If so, would it be 10.2?

Yes, likely. Yes, 10.2.
Although this is not a promise - 10.2 is still being planned.
But JSON has a good chance of ending up there.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Cross compile issue

[Sergei Golubchik]

Hi, Johnson!

On Sep 03, Johnson Cheng wrote:
> Dear Sergei,
> 
> Thanks for your reply.
> I am not sure what it happened because it is disappeared when I 
> re-cross-compile mariaDB at another new environment.
> Actually I am dig out a problem about DB_TXN error message as below,
>   [ 80%] Building CXX object 
> storage/tokudb/CMakeFiles/tokudb.dir/ha_tokudb.cc.o
>   In file included from 
> /home/johnson/SanSoft/model/Phoenix_130313-0901/Package/maria/mariadb-10.0.21/storage/tokudb/ha_tokudb.cc:122:0:
>   
> /home/johnson/SanSoft/model/Phoenix_130313-0901/Package/maria/mariadb-10.0.21/storage/tokudb/hatoku_defines.h:
>  In function int txn_begin(DB_ENV*, DB_TXN*, DB_TXN**, uint32_t, THD*):
>   
> /home/johnson/SanSoft/model/Phoenix_130313-0901/Package/maria/mariadb-10.0.21/storage/tokudb/hatoku_defines.h:407:19:
>  error: DB_TXN has no member named ‘set_client_id
>  this_txn->set_client_id(this_txn, thd_get_thread_id(thd));
>^
> Fortunately, I find out the root cause because I have installed BerkeleyDB at 
> first.
> It seems mariaDB and BerkeleyDB can't co-exist.

Yes, TokuDB uses uses data types with the same name.
But they can coexist (they do on my laptop), so that must be something
related to cross compilation again, may be cmake gets include paths
incorrectly...

> Now I can cross compile mariaDB every time, and I am doing the final fine 
> tune.

Congrats!

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Cross compile issue

[Sergei Golubchik]

Hi, Johnson!

On Sep 01, Johnson Cheng wrote:
> Dear Sergei,
> 
> I am sorry that I meet another cross compile issue about this,
> 
> Linking CXX shared module ha_archive.so
> [ 72%] Built target archive
> Scanning dependencies of target tokuportability_static_conv
> [ 72%] Building CXX object 
> storage/tokudb/ft-index/portability/CMakeFiles/tokuportability_static_conv.dir/huge_page_detection.cc.o
> In file included from 
> /home/johnson/SanSoft/model/Phoenix_130313-0831/build/sysroot/usr/include/features.h:378:0,
>  from 
> /home/johnson/SanSoft/model/Phoenix_130313-0831/build/sysroot/usr/include/sys/mman.h:22,
>  from 
> /home/johnson/SanSoft/model/Phoenix_130313/Package/mariadb/mariadb-10.0.21/storage/tokudb/ft-index/portability/huge_page_detection.cc:91:
> /home/johnson/SanSoft/model/Phoenix_130313-0831/build/sysroot/usr/include/sys/cdefs.h:286:63:
>  error: ’int stat(const char*, stat*)’ hides constructor for ’struct stat’ 
> [-Werror=shadow]
>  # define __nonnull(params) __attribute__ ((__nonnull__ params))
>^
> /home/johnson/SanSoft/model/Phoenix_130313-0831/build/sysroot/usr/include/sys/stat.h:219:6:
>  note: in expansion of macro ’__nonnull’
>   __nonnull ((1, 2));
>   ^
> /home/johnson/SanSoft/model/Phoenix_130313-0831/build/sysroot/usr/include/sys/cdefs.h:286:63:
>  error: ’int stat64(const char*, stat64*)’ hides constructor for ’struct 
> stat64’ [-Werror=shadow]
>  # define __nonnull(params) __attribute__ ((__nonnull__ params))
>^
> /home/johnson/SanSoft/model/Phoenix_130313-0831/build/sysroot/usr/include/sys/stat.h:229:47:
>  note: in expansion of macro ’__nonnull’
>   struct stat64 *__restrict __buf) __THROW __nonnull ((1, 2));
>^
> cc1plus: all warnings being treated as errors
> 
> I am not sure which library I miss. Could you give me some suggestions?

This looks like an issue with your system headers. TokuDB only makes it
fatal by using -Werror. If I were you I'd simply remove -Werror and
let the compilation continue.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Cross compile issue

[Sergei Golubchik]

Hi, Johnson!

On Aug 27, Johnson Cheng wrote:
 Dear Sergei,
 
 Sorry to update again. (I am a newer on cmake)
 I did some modifications on my configure command as below,
 LD=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-ld 
 NM=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-nm 
 CC=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-gcc 
 GCC=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-gcc
  
 CXX=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-g++
  
 RANLIB=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-ranlib
  \
 CFLAGS=-I/home/johnson/cross_prj/sysroot/usr/include 
 -I/home/johnson/cross_prj/sysroot/include \
 CXXFLAGS=-I/home/johnson/cross_prj/sysroot/usr/include 
 -I/home/johnson/cross_prj/sysroot/include \
 LDFLAGS=-L/home/johnson/cross_prj/sysroot/lib 
 -L/home/johnson/cross_prj/sysroot/usr/lib -lm -lz -ldl -lcrypt -pthread \
 CXXLDFLAGS=-L/home/johnson/cross_prj/sysroot/lib 
 -L/home/johnson/cross_prj/sysroot/usr/lib -lm -lz -ldl -lcrypt -pthread \
 ../../cmake/HOST-cmake/bin/cmake . \
 -DCMAKE_TOOLCHAIN_FILE=../cross.cmake \
 
 -DIMPORT_EXECUTABLES=/home/johnson/SanSoft/model/Phoenix_130313/Package/mariadb/native/import_executables.cmake
  \
 -DXTRADB_OK=1

This is not a good idea. I mean -DXTRADB_OK=1. XTRADB_OK is set
internally if HAVE_IB_GCC_ATOMIC_BUILTINS or HAVE_IB_SOLARIS_ATOMICS or
MSVC is defined. Depending on what exactly is defined, different parts
of the code are enabled. When you force XTRADB_OK=1 without enabling any
of the other defines, XtraDB will not enable any of these parts of the
code. Most probably it won't compile at all.

Instead of forcing -DXTRADB_OK=1, you need to do 
-DHAVE_IB_GCC_ATOMIC_BUILTINS=1.
Then XtraDB will compile as if gcc atomic builtins were available.
They probably are anyway, but cross-compilation caused the test to fail.

 It can success to configure, but it still has DB warning message. I am
 not sure if I can ignore it??
   CMake Warning at storage/tokudb/CMakeLists.txt:17 (MESSAGE):
 TokuDB is enabled, but jemalloc is not.  This configuration is not 
 supported

If you intend to use tokudb, then you probably should not ignore it.
if you don't intend to use tokudb, you can as well specify -DWITHOUT_TOKUDB=1
and it won't be compiled at all.

 Now I type make to do cross compiling, It seems it's not easy...
  [ 17%] Building CXX object 
 storage/xtradb/CMakeFiles/xtradb.dir/buf/buf0buddy.cc.o
  [ 17%] Building CXX object 
 storage/xtradb/CMakeFiles/xtradb.dir/buf/buf0buf.cc.o
  
 /home/johnson/SanSoft/model/Phoenix_130313/Package/mariadb/mariadb-10.0.21/storage/xtradb/buf/buf0buf.cc:
  In function �buf_block_t* buf_page_get_gen(ulint, ulint, ulint, ulint, 
 buf_block_t*, ulint, const char*, ulint, mtr_t*)’:
  
 /home/johnson/SanSoft/model/Phoenix_130313/Package/mariadb/mariadb-10.0.21/storage/xtradb/buf/buf0buf.cc:3021:55:
  error: �os_atomic_increment_ulint’ was not declared in this scope
os_atomic_increment_ulint(buf_pool-n_pend_unzip, 1);
  ^

That's exactly what I meant above. This error happens because you've
defined TOKUDB_OK=1

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Cross compile issue

[Sergei Golubchik]

Hi, Johnson!

On Aug 27, Johnson Cheng wrote:
 Dear Sergei,
 
 Thanks for your reply.
 I am sorry that I can NOT find any document about cross compile from 
 knowledge base.
 https://mariadb.com/kb/en/mariadb/getting-installing-and-upgrading-mariadb/

Yes, that's why I suggested to search the mailing list too. I wasn't
sure there's something in the knowledge base :(

 My host machine is Debian5. (gcc version is 4.3.2)
 My target machine is Intel x86_64 with gcc version is 4.8.3
 My Cmake version is 3.2.2 on host machine
 Here is my cmake command to configure mariadb-10.0.21 for cross compile,
   
 PATH=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/:/bin:/sbin:/usr/bin:/usr/sbin
  AR=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-ar 
 AS=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-as  
  LD=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-ld 
 NM=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-nm 
 CC=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-gcc 
 GCC=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-gcc
  
 CXX=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-g++
  
 RANLIB=/home/johnson/x86_64_TCglibc-2.19-3/usr/bin/x86_64-buildroot-linux-gnu-ranlib
  \
   CFLAGS=-I/home/johnson/ cross_prj /sysroot/usr/include -I/home/johnson/ 
 cross_prj /sysroot/include \
   CXXFLAGS=-I/home/johnson/ cross_prj /:sysroot/usr/include -I/home/johnson/ 
 cross_prj /sysroot/include \
   LDFLAGS=-L/home/johnson/cross_prj/sysroot/lib -L/home/johnson/ cross_prj 
 /sysroot/usr/lib -lm -lz -ldl -lcrypt -pthread \
   CXXLDFLAGS=-L/home/johnson/ cross_prj /sysroot/lib -L/home/johnson/ 
 cross_prj /sysroot/usr/lib -lm -lz -ldl -lcrypt -pthread \
   ../../cmake/HOST-cmake/bin/cmake .
 
 Unfortunately I meet the below error while typing make VERBOSE=1,
 [ 15%] Generating ../include/mysqld_error.h.tmp
 cd 
 /home/johnson/SanSoft/model/Phoenix_130313/Package/mariadb/mariadb-10.0.21/extra
   ./comp_err 
 --charset=/home/johnson/SanSoft/model/Phoenix_130313/Package/mariadb/mariadb-10.0.21/sql/share/charsets
  
 --out-dir=/home/johnson/SanSoft/model/Phoenix_130313/Package/mariadb/mariadb-10.0.21/sql/share/
  
 --header_file=/home/johnson/SanSoft/model/Phoenix_130313/Package/mariadb/mariadb-10.0.21/include/mysqld_error.h.tmp
  
 --name_file=/home/johnson/SanSoft/model/Phoenix_130313/Package/mariadb/mariadb-10.0.21/include/mysqld_ername.h.tmp
  
 --state_file=/home/johnson/SanSoft/model/Phoenix_130313/Package/mariadb/mariadb-10.0.21/include/sql_state.h.tmp
  
 --in_file=/home/johnson/SanSoft/model/Phoenix_130313/Package/mariadb/mariadb-10.0.21/sql/share/errmsg-utf8.txt
 ./comp_err: /lib/libc.so.6: version `GLIBC_2.14' not found (required by 
 ./comp_err)
 ./comp_err: /lib/libc.so.6: version `GLIBC_2.17' not found (required by 
 ./comp_err)
 make[2]: *** [include/mysqld_error.h.tmp] Error 1
 
 It tries to use my cross compiled extra/comp_err utility to do something at 
 my host machine.

Right...
During the build few helper tools are compiled and used in the build.
See http://www.cmake.org/Wiki/CMake_Cross_Compiling
In particular, we implement the solution from
http://www.cmake.org/Wiki/CMake_Cross_Compiling#Using_executables_in_the_build_created_during_the_build

So, you need to build for the native platform first. Not everything, just
import_executables target:

  make import_executables

This will create import_executables.cmake file. When you cross-compile
you'll need to set IMPORT_EXECUTABLES to the path to that file.
 
Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Cross compile issue

[Sergei Golubchik]

Hi, Johnson!

On Aug 25, Johnson Cheng wrote:
 Hi,
 
 I have problem to do cross compile on mariadb-10.0.21.
 Where can I find the relative document?

Check the knowledgebase and if there's nothing there, then mailing list
archives. I'm not sure about knowledgebase, but this topic was certainly
discussed on the mailing list.

What is the problem that you have?

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Latest round of Oracle CVE status on MariaDB

[Sergei Golubchik]

Hi, Brian!

Thanks. I've updated the security page now.
I think that CVE-2015-4757 is fixed in 5.5.43 (and 10.0.18), and
  CVE-2015-4752
  CVE-2015-2648
  CVE-2015-2643 
  CVE-2015-2582
are fixed in 5.5.44 (and 10.0.20).

While I cannot be sure what CVE-2015-4737 and CVE-2015-2620 are about,
I suspect that the first is
https://github.com/mysql/mysql-server/commit/c655515d and the second is
https://github.com/mysql/mysql-server/commit/fdae90dd.

If that's right, than the first is intentional behavior, not a bug.
I believe that changing it might break user applications (esp. backups).

The second isn't a fix it all, it only covers one very specific case.
I've created MDEV-8269 to have this bug properly fixed.

Regards,
Sergei

On Aug 13, Brian Evans wrote:
 The quarterly CVE list from oracle was published[1].
 The following CVEs are listed there affecting 5.5, but not listed on
 the MariaDB security page[2].
 
 CVE-2015-4757
 CVE-2015-4752
 CVE-2015-2648
 CVE-2015-2643
 CVE-2015-2582
 CVE-2015-4737
 CVE-2015-2620
 
 Are they fixed with 5.5.45 and 10.0.21, or any other version?
 
 Brian
 
 [1] 
 http://www.oracle.com/technetwork/topics/security/cpujul2015-2367936.html#AppendixMSQL
 [2] https://mariadb.com/kb/en/mariadb/security/
 

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] why stdout? - [Note] /usr/libexec/mysqld (mysqld 10.0.20-MariaDB) starting as process 115096

[Sergei Golubchik]

Hi, Reindl!

On Aug 06, Reindl Harald wrote:
 Aug  6 12:24:24 testserver mysqld: 150806 12:24:24 [Note] 
 /usr/libexec/mysqld (mysqld 10.0.20-MariaDB) starting as process 115096 ...
 
 why is this now stdout instead mysql logs?

I cannot repeat it.
Here:

$ ./mysqld out 2err 
$ grep starting err out
err:150806 13:53:24 [Note] /home/serg/maria/sql/mysqld (mysqld 
10.0.21-MariaDB-debug) starting as process 22001 ...
^^^

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] why stdout? - [Note] /usr/libexec/mysqld (mysqld 10.0.20-MariaDB) starting as process 115096

[Sergei Golubchik]

Hi, Reindl!

On Aug 06, Reindl Harald wrote:
 Am 06.08.2015 um 14:25 schrieb Sergei Golubchik:
  On Aug 06, Reindl Harald wrote:
  Am 06.08.2015 um 13:56 schrieb Sergei Golubchik:
  On Aug 06, Reindl Harald wrote:
  Aug  6 12:24:24 testserver mysqld: 150806 12:24:24 [Note]
  /usr/libexec/mysqld (mysqld 10.0.20-MariaDB) starting as process 115096 
  ...
 
 the point is that echo something to stdout/stderr in case of a
 background service burries the lines and the only reason they appear
 at all is journald collecting that stuff and forwards it to syslog
 
 they should go to log-error as all other messages

Ah, okay. You're right, this line is sent to stderr before stderr is
redirected to log-error. I suppose it's a bug.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] will maria db overcome mysqls feature limitations? (foreign keys+triggers, recursive triggers)

[Sergei Golubchik]

Hi, Ludwig!

On Jun 21, Ludwig Gramberg wrote:
 Hi,
 
 I was wondering if mariadb plans to overcome mysqls limitations such as
 - foreign key operations do not start triggers
 - triggers can’t be recursive (error 1442)

I haven't heard about any plans about these specific limitations.
That's because nobody requested that yet.

But by all means, do request these feature requests at mariadb.org/jira
and then we will certainly consider them for the next major version (and
then for the following major version if they won't made it into the
next, and so on).

 But besides additional features, is there a chance that mariadb could
 ever overcome limitations?  Is this even a goal of this project?

Yes, sure. But we cannot do everything at once, so we prioritize
tasks that our users wants us to do.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] will maria db overcome mysqls feature limitations? (foreign keys+triggers, recursive triggers)

[Sergei Golubchik]

Hi, Justin!

On Jul 19, Justin Swanhart wrote:
 Hi,
 
 Foreign keys need support in SE to fix the first item.  Cascading
 operations don't work for triggers, but even worse, they don't work
 for RBR replication if a) storage engines differ on master and slave
 or b) the slave lacks the constraints.  This is because InnoDB does
 cascade operations internally, because the SE interface doesn't
 support foreign keys, thus the cascade operations only work on the
 slave if the SE matches and the constraints match.  This basically is
 a RBR/pseudo-SBR mix that simply doesn't work.

Right. My first idea was to add some kind of a callback for a storage
engine to inform the server about cascading operation. Then it could
fire a trigger.

We already can fire triggers for low-level row operations, this is used
in our RBR triggers feature:
https://mariadb.com/kb/en/mariadb/running-triggers-on-the-slave-for-row-based-events/

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.1.4 Progress output

[Sergei Golubchik]

Hi, Rhys!

On Apr 20, Rhys Campbell wrote:
 Note the progress output below. I've seen several like this. i.e...
 
 Stage: 1 of 72 'altering table' 350% of stage done. Is this intentional?

No, not exactly :)

This happens when the engine miscalculates number of steps it needs to
do for an ALTER. If you have a test case for that behaviour, we'll fix
it.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] performance_schema

[Sergei Golubchik]

Hi, Rhys!

On May 14, Rhys Campbell wrote:
 Hi All,
 
 Does anyone know if the performance_schema guarantees that every
 statement executed will appear in the
 events_statement_summary_by_digest table (excluding those that could
 be pushed out due to size restrictions). Or is it a best effort
 system?

It is supposed to guarantee that. Do you have a test case where it
misses some executed statements?

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] query cache dump_at_shutdown/load_at_startup useful + no character set/collation?

[Sergei Golubchik]

Hi, Daniel!

On May 12, Daniel Black wrote:
 I was testing out using the query_cache_info plugin 
 (https://mariadb.com/kb/en/mariadb/query_cache_info-plugin/) to pre-cache 
 data into a server after an upgrade. The read only slaves have a fairly high 
 cache hit ratio.
 
 select concat('use ', STATEMENT_SCHEMA, ';', STATEMENT_TEXT, ';') as q INTO 
 OUTFILE '/tmp/qc.txt' FIELDS ESCAPED BY '' TERMINATED BY '' from 
 information_schema.QUERY_CACHE_INFO;
 Query OK, 7122 rows affected (0.03 sec)
 
 When loading these it seems the statements have a collation however this 
 isn't stored in the query cache.
 
 $ mysql --force  /tmp/qc.txt  /dev/null
 ERROR 1267 (HY000) at line 119: Illegal mix of collations 
 (latin1_bin,IMPLICIT) and (utf8_general_ci,COERCIBLE) for operation '='
 ERROR 1267 (HY000) at line 2390: Illegal mix of collations 
 (latin1_bin,IMPLICIT) and (utf8_general_ci,COERCIBLE) for operation '='
 ERROR 1267 (HY000) at line 7569: Illegal mix of collations 
 (latin1_bin,IMPLICIT) and (utf8_general_ci,COERCIBLE) for operation '='
 
 Attempts to force this with   --default-character-set={utf8|utf8mb4} also 
 resulted in the same error or same with (echo set 
 character_set_connection=utf8;; cat /tmp/qc.txt; ) | time mysql --force..
 
 Are query cache results actually predictable if they have the same
 statement running in a different session character set/collation?

Yes, of course.
See in sql_cache.{h,cc} the structure Query_cache_query_flags and what's
in there.

But query_cache_info plugin doesn't show flags.

 Anyone see a work around to get the restore file usable on all
 queries?

Perhaps, extending query_cache_info plugin to show flags?
Your script will  need to restore a lot more than just a collation.
Some flags cannot be set from the sql script whatsoever - but you can
skip these entries in your select concat query.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.18 now available

[Sergei Golubchik]

Hi, Reindl!

On May 07, Reindl Harald wrote:
 
  No, it affects the server, not mysql_upgrade. But it's a new
  statement, that mysql_upgrade is using,  no existing query can
  possibly trigger that bug
 
 well, in other words anybody could crash the server by write a
 specific query and so i am not sure what is worser: the security bugs
 in 10.0.17 or that bug in 10.0.18

Right. We'll release 10.0.19 to fix that.

 doesn't upstream run mysql_upgrade mandatory independent of changes?

No. Depends on what upstream is. Debian/Ubuntu do that, as far as I
remember. RedHat/Fedora/CentoS - don't (again, as far as I remember).

 OpenVAS against 10.0.17 reports CVE-2013-1861 and CVE-2012-5627 while 
 there still was no answer to the mail below and so the state which of 
 the mysql security bugs are also in mariadb is unknown

I've updated MariaDB.org CVE overview page about a week ago.
(note that email didn't request an answer, it requested the page to be
updated)

Regards,
Sergei

  Weitergeleitete Nachricht 
 Betreff: [Maria-developers] Oracle April security notices and MariaDB
 Datum: Sun, 19 Apr 2015 21:55:19 +0300
 Von: Otto Kekäläinen o...@seravo.fi
 An: maria-develop...@lists.launchpad.net 
 maria-develop...@lists.launchpad.net
 
 Hello!
 
 Debian security team is pressing me on the information about which
 recent Oracle CVEs affect MariaDB and which not. They default to
 assuming all affect so we need to prove otherwise.
 
 The Debian CVE tracker:
 https://security-tracker.debian.org/tracker/source-package/mariadb-10.0
 
 None of these recent CVEs are listed at the MariaDB.org tracker:
 https://mariadb.com/kb/en/mariadb/security/
 
 Could somebody please update the MariaDB.org CVE overview page?

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.18 now available

[Sergei Golubchik]

Hi, Reindl!

On May 07, Reindl Harald wrote:
 Am 07.05.2015 um 21:27 schrieb Sergei Golubchik:
  Hi, Reindl!
 
  On May 07, Reindl Harald wrote:
  Don't build on Fedora 21 / gcc-4.9.2-6.fc21.x86_64 but on Fedora 20 /
  gcc-4.8.3-7.fc20.x86_64 which is not a good sign since 10.0.17 with
  identical environment builds on both
 
  This doesn't help much, because the actual error was somewhere above
  the line you've started your paste from.
 
 i had attached the mentioned /tmp/cctR4nzQ.out
 anyways, complete buildlog (stdout and stderr) attached

Here are relevant lines:

Building C object strings/CMakeFiles/strings.dir/ctype-mb.c.o
cd /home/builduser/rpmbuild/BUILD/mariadb-10.0.18/strings  /usr/bin/cc 
-DDISABLE_MYSQL_THREAD_H -DHAVE_CONFIG_H -m64 -O2 -march=sandybridge 
-mtune=sandybridge -fopenmp -mmmx -msse2 -msse3 -msse4.1 -msse4.2 -mavx 
-mpclmul -mcrc32 -maes -mfpmath=sse -pipe -fomit-frame-pointer 
-finline-functions -finline-limit=60 -fexceptions -fstack-protector-strong 
--param=ssp-buffer-size=6 -D_FORTIFY_SOURCE=2 -Wstack-protector -Wformat 
-Werror=format-security -O3 -fno-stack-protector -fstack-protector 
--param=ssp-buffer-size=8 -fPIC -fomit-frame-pointer -fno-exceptions 
-ffixed-ebp -fwrapv -fno-strict-aliasing -fno-delete-null-pointer-checks 
-D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE  -fPIC -O3 -g 
-static-libgcc -fno-omit-frame-pointer -fno-strict-aliasing  -Wno-uninitialized 
-DDBUG_OFF -DMY_PTHREAD_FASTMUTEX=1 
-I/home/builduser/rpmbuild/BUILD/mariadb-10.0.18/include-o 
CMakeFiles/strings.dir/ctype-mb.c.o   -c 
/home/builduser/rpmbuild/BUILD/mariadb-10.0.18/strings/ctype-mb.c
[  7%] /home/builduser/rpmbuild/BUILD/mariadb-10.0.18/strings/ctype-mb.c: In 
function 'my_like_range_mb':
/home/builduser/rpmbuild/BUILD/mariadb-10.0.18/strings/ctype-mb.c:905:1: 
internal compiler error: in dwarf2out_frame_debug_expr, at dwarf2cfi.c:1550
 }
 ^
Please submit a full bug report,
with preprocessed source if appropriate.
See http://bugzilla.redhat.com/bugzilla for instructions.

The same error happens in six files:
ctype-mb.c, ctype-uca.c, ctype-simple.c, dtoa.c, decimal.c, ctype-ucs2.c

Between 10.0.17 and 10.0.18 there were, indeed, changes in strings/ctype-uca.c
and in strings/decimal.c. But other files were *not* changed at all.

So I suspect it's something local to your system, like disk full or gcc
bug. Only these six files are affected, because make runs six
compilations in parallel and stops at the first failure, so you cannot
see more than six failures from it.

 attached - what makes me really worry is that the build suceeds on 
 Fedora 20 but crashs due myql_upgrade the mysqld process, see other mail

Yes, I replied to that mail already. That was a pretty bad bug, and
unfortunately discovered too late :(

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.18 now available

[Sergei Golubchik]

Hi, Reindl!

On May 07, Reindl Harald wrote:
 Am 07.05.2015 um 22:29 schrieb Sergei Golubchik:
  On May 07, Reindl Harald wrote:
 
  Am 07.05.2015 um 21:02 schrieb Reindl Harald:
  Don't build on Fedora 21 / gcc-4.9.2-6.fc21.x86_64 but on Fedora 20 /
  gcc-4.8.3-7.fc20.x86_64 which is not a good sign since 10.0.17 with
  identical environment builds on both
 
  and the Fedora 20 build is completly broken and crashs due mysql_upgrade
 
  Yes, I've just fixed it. But too late for 10.0.18 :(
 
  Your most simple workaround would be not to run mysql_upgrade. It
  wouldn't do anything noticeable when upgrading from 10.0.17 to 10.0.18
  anyway
 
 does that *really* only affect mysql_upgrade since a crash of mysqld 
 itself and not mysql_upgrade is bad and i am not sure if some other 
 query in production would not trigger the same problem?

No, it affects the server, not mysql_upgrade. But it's a new statement,
that mysql_upgrade is using,  no existing query can possibly trigger
that bug.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.18 now available

[Sergei Golubchik]

Hi, Reindl!

On May 07, Reindl Harald wrote:
 
 
 Am 07.05.2015 um 21:02 schrieb Reindl Harald:
  Don't build on Fedora 21 / gcc-4.9.2-6.fc21.x86_64 but on Fedora 20 /
  gcc-4.8.3-7.fc20.x86_64 which is not a good sign since 10.0.17 with
  identical environment builds on both
 
 and the Fedora 20 build is completly broken and crashs due mysql_upgrade

Yes, I've just fixed it. But too late for 10.0.18 :(

Your most simple workaround would be not to run mysql_upgrade. It
wouldn't do anything noticeable when upgrading from 10.0.17 to 10.0.18
anyway.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.18 now available

[Sergei Golubchik]

Hi, Reindl!

On May 07, Reindl Harald wrote:
 
 interesting - in fact it's -mtune=sandybridge versus -mtune=westmere 
 while i am pretty sure that i rebuilt any local maintained packages 
 recently for sandybrige to test because a hardware replacement is 
 planned in august resulting in the oldest CPU to support is sandybrige based
 
 i tried a ton of other packages on the recent environment with all 
 updates applied and mariadb is the only package failing to build while 
 httpd, php, apr, openvpn, openssl, ffmpeg are still fine
 
 https://bugzilla.redhat.com/show_bug.cgi?id=1219648

Well, I'm not saying MariaDB should be used as a compiler testing tool,
but here, https://bugzilla.redhat.com/show_bug.cgi?id=1149660 a bug in
fedora 21 binutils, also found when compiling mariadb :)

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Password Complexity Plugin

[Sergei Golubchik]

Hi, Adam!

On Apr 27, Adam Scott wrote:
 It's taken some time, but I have backported the simple_password_check
 plugin and the password api to MariaDB 5.5  The mysql-test is
 succeeding on all but the test of the strict_password_validation
 variable.
 
 My question is, should a pure backport include support for
 strict_password_validation variable (
 https://mariadb.com/kb/en/mariadb/server-system-variables/#strict_password_validation)
 ?  I suspect yes, but wanted to ask to start a further discussion:  namely
 if I submit this patch will it be included in future MariaDB 5.5 releases?

I believe it's unlikely. If we wanted this feature in 5.5, it would've
been implemented in 5.5 in the first place.

But 5.5 is already GA, it was GA for quite a while. We don't add new
feature to GA versions, that's why password validation was added to
10.1.

I think that if you'll keep your own version of 5.5 + password
validation and use that in your environment - it should not require much
efforts to maintain. 5.5 code base is very stable at the moment, changes
are few and sql_acl.cc almost doesn't change at all. And then you can
skip 10.0 and upgrade directly to 10.1 one day :)

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] user map auth plugin

[Sergei Golubchik]

Hi, Felipe!

On Apr 28, Felipe Gasper wrote:
 So, I recently wrote an auth plugin with the intent of implementing a
 simple user map, authenticated against the hashed password that would
 be retrieved via the normal password entry scheme.
 
 I got far too far along before realizing that this scheme basically
 requires that either the client use a plugin to communicate with the
 server auth plugin, or I use the old MySQL auth plugin.

yes :(
the server is hard-coded to use built-in plugins when the standard mysql
authentication is used.

 ...or is there some way to re-initialize secure MySQL auth from the
 server auth plugin??

I think so, yes. Your plugin should to declare that it needs
mysql_native_password plugin on the client side.

Then the server will notice that the client already uses
mysql_native_password plugin and will only reinitialize the server part
of the authentication without client noticing anything.

Tell me if it'll work :)

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Proxy users in MariaDB?

[Sergei Golubchik]

Hi, Felipe!

On Apr 24, Felipe Gasper wrote:
 On 4/23/15 5:48 PM, Geoff Montee wrote:
 
  I'm not a big fan of this bit from the MySQL documentation:
 
  When a single account has been granted proxy privileges on more than
  one account, the server mapping is nondeterministic. Therefore,
  granting proxy privileges on multiple accounts to a single account is
  discouraged.
 
  Nondeterministic behavior can be pretty messy. Maybe improving the
  role system to support more use cases would be better than going down
  this route?
 
 Agreed. It should fail, IMO, when you try to add a 2nd PROXY privilege 
 to the same user. Very strange design.

That's because they're abusing the PROXY privilege.

PROXY privilege is exactly that, a *privilege* - it allows a given user
to pretend that he's some other given user. What user should it be -
this is defined by a plugin. A plugin says let's Alice be Bob for this
session. And DBMS checks whether Alice is allowed to be Bob. And,
naturally, there can be many proxies granted to the same user. Alice
might also be allowed to be Carol and Dave, but not Eve.

PROXY privilege was never supposed to be an *instruction* of what a user
will be. When you grant a SELECT privilege on a table to someone, this
someone doesn't get the table dumped out to him every time he connects,
does he? :) A privilege is not an instruction, it doesn't tell what to
do, only what you can do.

  It would be nice to have the flexibility to allow roles to log in
  (similar to how PostgreSQL roles can be defined with WITH LOGIN role
  attributes), but I'm not sure if MariaDB will get that feature. Maybe
  submit a feature request to our JIRA?
 
 Done: https://mariadb.atlassian.net/browse/MDEV-8047
 
 I’m not sure it’s filed in quite the best way (e.g., it didn’t let me 
 select “improvement” as the type)

You've done it correctly, the type Task is what it should be.

 I wonder what the perceived advantage was/is of keeping users and roles 
 as separate concepts.

I think that's what SQL standard says.

We've implemented roles in 10.0 looking into the standard and following it
almost up to the letter.

In 10.1 we've added an extension - default roles.
We can add another extention, that's surely possible.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] How many open files by daemon is still sane?

[Sergei Golubchik]

Hi, Justin!

On Apr 21, Justin Swanhart wrote:
 
 It is in fact, negatively scaleable without partitioning it:
 http://www.percona.com/blog/2009/11/16/table_cache-negative-scalability/

This doesn't directly apply to MariaDB. We didn't partition it
because our table definition cache is lock-free. There were quite a few
related changes in 10.0 (e.g. see MDEV-7292 and linked issues). In
short, we didn't partition it, because it doesn't need to be
partitioned. Not for this benchmark workload, at least.

Regards,
Sergei

  I think original question was about 5.5.
 
  MySQL 5.6 has partitioned table cache, but rather to overcome the
  negative scalability aspect of increasing number of concurrent
  connections.
 
  No version of MariaDB has partitioned table cache. At least yet.
 

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] SHOW GRANTS change?

[Sergei Golubchik]

Hi, Felipe!

On Apr 13, Felipe Gasper wrote:
 Note: I believe “dbguy” came from a My5.6 - Ma10.0 upgrade.
 
 Is it possible that something in MariaDB isn’t reading MySQL stuff 
 correctly?

Could be. There was one issue in the way MySQL started confusing a
password and the authentication string for plugins.

Could you show the complete line for that user?
select * from mysql.user where user='dbguy' \G

Thanks!

Regards,
Sergei

 -FG
 
 On 13 Apr 2015 2:14 PM, Felipe Gasper wrote:
  Dear Guillaume,
 
   This is what I see:
 
  
  MariaDB [(none)] select host, password from mysql.user where user='dbguy';
  +---+---+
  | host  | password  |
  +---+---+
  | localhost | *3D4FD9A495C3E81883E1A42AD351871F74F7CAA9 |
  +---+---+
  1 row in set (0.00 sec)
 
  MariaDB [(none)] show grants for 'dbguy'@'localhost';
  ++
  | Grants for dbguy@localhost |
  ++
  | GRANT USAGE ON *.* TO 'dbguy'@'localhost'  |
  | GRANT ALL PRIVILEGES ON `dbguy\_goal`.* TO 'dbguy'@'localhost' |
  ++
  2 rows in set (0.00 sec)
 
  MariaDB [(none)] select user();
  ++
  | user() |
  ++
  | root@localhost |
  ++
  1 row in set (0.00 sec)
  ---
 
   That hash is the hash of the string '234wer'.
 
   There does seem to be a bug here … ?
 
  -FG

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] [Maria-developers] Enabling feedback pluging for MariaDB 10.1.4

[Sergei Golubchik]

Hi, Ian!

On Mar 11, Ian Gilfillan wrote:
 I understand that the statistics would have some meaning, and
 understand that there's no actual data being sent, etc.
 
 However, enabling the feedback plugin by default IS going to create a
 whole lot of negative publicity for MariaDB. Is the limited data the
 feedback plugin provides worthwhile? Look at the similar situation
 with Ubuntu/Canonical. No matter how much they argued there was no
 real risk, they suffered a huge amount of negative publicity and
 eventually reversed the decision.

Yes. This is my worry too. That's why Monty has sent this email,
that's why it's planned to be enabled only for beta.

But asking people to please enable the feedback plugin just doesn't
work. It seems that most of our users don't care about it at all - I
mean, they don't care whether it's enabled or not, and don't care to do
anything.  That is, if the plugin is disabled - it'll stay disabled,
that's what is happening on Linux. If it's enabled - it'll stay enabled,
that's what is happening on Windows (MariaDB Windows installer has a
special dialog about it with a checkbox that's enabled by default, if
I'm not mistaken).

 Why go through this pain? Rather use this as an opportunity to create
 a thorough survey (which can provide far more detailed and useful
 information than the plugin running on a beta version) and engage
 constructively with the community. The survey has other benefits too,
 as pointed out elsewhere in the thread - raising awareness of
 lesser-known features, and just spreading the survey will be marketing
 for MariaDB.

Yes, survey is a good idea and I agree that we should do it.

 And by all means, give people an option to enable the plugin on 
 installation, as long as it defaults to off.

Okay, I'll see how we could pop up a dialog box during installation
to give users an option to enable or disable the plugin. With the plugin
being automatically disabled in non-interactive installations.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] [Maria-developers] Enabling feedback pluging for MariaDB 10.1.4

[Sergei Golubchik]

Hi, Jeremy!

On Mar 12, Jeremy Cole wrote:
 
  But asking people to please enable the feedback plugin just
  doesn't work. It seems that most of our users don't care about it at
  all - I mean, they don't care whether it's enabled or not, and don't
  care to do anything.  That is, if the plugin is disabled - it'll
  stay disabled, that's what is happening on Linux. If it's enabled -
  it'll stay enabled, that's what is happening on Windows (MariaDB
  Windows installer has a special dialog about it with a checkbox
  that's enabled by default, if I'm not mistaken).
 
 Hmm, I would seriously wonder whether installations that use the
 default configuration and don't change/manage it will be the best
 source of data about customized configuration. (Or anything, really.)
 
 Is it really that useful to get a bunch of stats about packages that
 were installed perhaps only to meet some prerequisite for another
 package?

Yes, of course. Stats will show what features this another package uses
(xml functions, gis, time zones, plugins, etc) and this is precisely
what we would like to know. It doesn't matter whether the configuration
is customized or default.

But if the mariadb is pulled in as a dependency, but the server is not
started then it won't send any data - and, again, that's correct, it
doesn't count as usage if it doesn't run.

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] [Maria-developers] Enabling feedback pluging for MariaDB 10.1.4

[Sergei Golubchik]

Hi, Vladislav!

On Mar 12, Vladislav Vaintroub wrote:
 
 Hi Serg,
 On 03/12/15 09:17, Sergei Golubchik wrote:
  MariaDB Windows installer has a special dialog about it with a 
  checkbox that's enabled by default, if I'm not mistaken).
 
 I'm sorry, but you are mistaken :)  Checkbox is off by default.

Okay :) I did look at the feedback dialog in extra.wxs, but I'm not very
surprised that I got it wrong.

 Also, at some point in the past I  compared the download number of MSIs 
 with the number that shows up in feedback plugin stats. The later number 
 (# of enabled feedback plugin) is much less, by 2 orders of magnitude. 

Still, 95% of reports are coming from Windows. Just having a checkbox
it a tremendous help, even if it's disabled by default.

 People seem to  be happy with default installer settings, in general.

My point, exactly.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] R: GSOC 2015 :UDFs returning arrays

[Sergei Golubchik]

Hi, Justin!

On Mar 10, Justin Swanhart wrote:
 wl#820 also has table functions in it:
 http://antbits.blogspot.com/2009/01/table-functions-in-mysql.html
 
 mysql INSTALL PLUGIN Deep_Thought SONAME 'psm_example.so';
 mysql CREATE FUNCTION test.FooAnswer()
 - RETURNS TABLE(answer TEXT)
 - NO SQL LANGUAGE Deep_Thought EXTERNAL NAME 'compute';
 mysql CREATE VIEW foobar AS SELECT * FROM TABLE(test.FooAnswer) AS wibble;
 mysql SELECT * FROM foobar WHERE answer LIKE 'F%';
 +---+
 | answer|
 +---+
 | Forty-Two |
 +---+
 
Ah, indeed. Thanks.

It's still an unrelated feature - I mean, there is no reason why a table
function must necessarily be impemented as an external procedure.

And the implementation - I believe - creates a temporary table and puts
all the data into it. I would rather prefer a generator-like interface,
that produces rows one by one. It works much better for big data sets.
Or, may be, we can do both.

Either way - this implementation can be very helpful. But it'll need at
least to be updated to work in 10.1. And may be we add the generator
interface.

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Enabling feedback pluging for MariaDB 10.1.4

[Sergei Golubchik]

Hi, Justin!

On Mar 09, Justin Swanhart wrote:
 
 I agree with Kristian.  Given the way it works, the statistics are
 really meaningless and I feel you shouldn't drive important choices
 based on bad statistics.

Of course. This statistics is not *the only* argument. For important
decisions there are always many aspects to consider. The statistics is
just another data, in addition to and *I* think that every user needs
feature X and never uses the feature Y :)

 I personally would suggest displaying a link to a feedback/survey form
 with web downloads and display a message after rpm/deb installation
 that says something like please visit http://blah/blah/blah/survey to
 tell us more about the features you use and help direct the future
 development of MariaDB.  This has an added bonus: not all users know
 about all features, and a list/survey of the important and interesting
 ones could get more users to use them.

A survey is a pretty good idea, thanks!

It may not provide a much better (as in representative sample) statistics,
but it will surely tell the users about the features.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Enabling feedback pluging for MariaDB 10.1.4

[Sergei Golubchik]

Hi, Kristian!

On Mar 09, Kristian Nielsen wrote:
 
  As most MariaDB users should know, the feedback is totally anonymous
  and no private or sensitive information is being sent.
 
  Any comments, suggestions or recommendations?
 
 I think it is a bad idea. Please do not do it.
 
 Phone-home is a misfeature in any product, and even more so in system
 software like a database.

Agree. I don't like it myself. Still between that and random development
decisions driven by the marketing department, I just might prefer
phone-home.

 And besides, the information is much less useful than you think,
 because of unknown, but probably extreme, data skew. In fact, it will
 probably be more harmful than useful because people will use bad data
 to justify bad decisions.

Right, but there are three approaches to this. First, try to get more
reports, in the hope that it'll be a representative sample. That's what
we're discussing this email thread. Second, take the known skew into
account when analyzing the data. For example, see the OS stats chart
(http://mariadb.org/feedback_plugin/stats/os/) - it doesn't mean that
96.3% of MariaDB installations are on Windows, it means that we have
disproportionally more reports from Windows. And third, use numbers
where the skew doesn't matter. For example, total number of
installations.

 Experience supports this point of view with our download numbers. They
 do not include apt-get / yum / etc. installations, which judging from
 IRC conversations are the majority. Yet people continuely refer to
 them as though they mean anything, just because they are there.

Of course they mean something. They show, literally, how many times
mariadb was downloaded from downloads.mariadb.org. As such, they show
that mariadb was downloaded (from all sources) *at least* that many
times.

Also, one can *reasonably assume* that the number of downloads from all
other sources follows the grows of downloads from mariadb.org (unless
the number of other sources changes).

These numbers mean a lot. They just don't mean the total number of all
mariadb installation (from all sources). Which is pretty obvious :)

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Enabling feedback pluging for MariaDB 10.1.4

[Sergei Golubchik]

Hi, Adam!

On Mar 09, Adam Scott wrote:
 Maybe make it an option when installing?

Yes, that'd be great.

On Windows there's a GUI installer, and it has a checkbox for feedback
plugin. That's why we get 95% of reports from Windows. Most users don't
mind having it enabled, so it seems.

So if rpm/deb packages would ask about feedback - that'll help a lot. It
just needs be done carefully not to break unattended installs.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Enabling feedback pluging for MariaDB 10.1.4

[Sergei Golubchik]

Hi, Justin!

On Mar 10, Justin Swanhart wrote:
 
 Where is it described exactly what is collected?  Descriptions I see
 say it is basically ...  well, no, I want a full description of all
 data collected, particularly if it collects versions of software as
 knowing what version of software I'm running lets you know what I'm
 vulnerable to.

It's explained here:  https://mariadb.com/kb/en/feedback-plugin/
Basically :) you can do

  mysql -e 'select * from information_schema.feedback'  report.txt
  curl -F data=@report.txt https://mariadb.org/feedback_plugin/post

and the result will be exactly the same. And you can set --feedback-url
to any url of your choice and see exactly what is being sent.

 Is the data sent via SSL?

Yes, by default. Unless you change feedback_url to use http, not https.

 Is the data stored encrypted in your data center?

No, I don't think so.

 I certainly don't want my c library version, mariadb version, etc,
 sent in clear over the internet where anybody can read it, and I don't
 want it stored unencrypted at rest somewhere, where someone can just
 abscond with it.

C library version is not sent, MariaDB version is. But they're not tied
to you - nobody can trace these data back, we certainly cannot.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Enabling feedback pluging for MariaDB 10.1.4

[Sergei Golubchik]

Hi, Federico!

On Mar 10, Federico Razzoli wrote:
 I am no lawyer, but please consider possible legal problems for users.
 1) I sign an NDA with my customer
 2) I enable Feedback
 3) You see things I shouldn't reveal.
 I know that my data are not sent. And you say it's anonymous.
 But you will have at least the server's IP and MAC.
 The real problem is not if we trust trusting YOU (I do) - the real
 problem is that sending that data could be illegal.

Right, this could happen.
I have no solution for this, only few assorted thoughts:

* not we say it's anonymous, you don't have to trust, you can verify
  it. It's not a solution, because under NDA you might not be allowed to
  send even anonymous data. Still you can see what is being sent, no
  need to trust.

* We won't have MAC address. But the IP address will be in the apache
  logs (even if I'd say we won't log IP addresses you won't be able to
  verify it).

* The idea was to enable feedback plugin in beta and disable it before
  GA. And beta versions come with a warning don't use in production.

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] stored programs

[Sergei Golubchik]

Hi, Justin!

On Mar 03, Justin Swanhart wrote:
 So uhmm, use a new parser.  Like the one antony wrote in LLVM? :)  I
 mean, we would like to have window functions, CTE, and all those other
 fun things. SQL-92 is a not relevant anymore.

In MySQL times we've tried to replace a parser. Twice.
Antony has spent a year on that - that was one of the attempts.

We're still using the old one, so neither project was a success.

And Oracle started parser refactoring in 5.7, but they're doing it
piece-wise now, not trying to replace everything at once. And they'll
have to stay with bison this way.

What I mean - we very much want to use a new parser. We really do.
But it's just such a huge project :(

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] stored programs

[Sergei Golubchik]

Hi, Igor!

On Mar 03, Igor Babaev wrote:
  
  I'd also like to discuss window functions too.  I've implemented them
  in shard-query and have ideas about how to implement them in the
  server, but pluggable parser would be really useful here.
  
  Window functions have a good chance of being in 10.2, it's MDEV-6115.
  But I don't think that somebody is working on MDEV-6115 yet.
 
 I started working on MDEV-6115 some time ago.

Ah, great. Sorry, I didn't know it.
Could you then discuss it with Justin, please?
See above, he has some ideas about the implementation.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Giving up on dynamic columns

[Sergei Golubchik]

Hi, Tom!

On Mar 01, Tom Worster wrote:
 so it turns out the problem is my incompetence. i thought blob allowed
 long values.
 
 i've been using mysql since 3.something and i still don't know much
 about its abundance of types.
 
 i retract my comments and crawl back into my hole.

In fact, I found your bug report very helpful. The fact that
COLUMN_CHECK does not always detect data truncation is a bug.
Thanks for reporting this behavior!

We'll try to fix it.

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Dependency questions regarding MariaDB 10 on CentOS6.5 with local mirror of repo

[Sergei Golubchik]

Hi, Sophie!

On Feb 23, Sophie L wrote:
 Hi everybody,
 
   I mirrored MariaDB repo,but noticed servers had unmet dependencies when
 I tried to install.  I have this message:
 
 Error: Package: MariaDB-client-10.0.16-1.fc20.x86_64 (mariadb)
 Error: Package: MariaDB-client-10.0.16-1.fc20.x86_64 (mariadb)
Requires: libstdc++.so.6(GLIBCXX_3.4.15)(64bit)
 Error: Package: galera-25.3.5-1.sles12.x86_64 (mariadb)
Requires: libssl.so.1.0.0()(64bit)
 Error: Package: galera-25.3.5-1.sles12.x86_64 (mariadb)
Requires: libcrypto.so.1.0.0()(64bit)
 Error: Package: MariaDB-Galera-server-10.0.16-1.fc20.x86_64 (mariadb)
Requires: libstdc++.so.6(GLIBCXX_3.4.15)(64bit)
 Error: Package: MariaDB-client-10.0.16-1.fc20.x86_64 (mariadb)
Requires: libc.so.6(GLIBC_2.17)(64bit)
 Error: Package: MariaDB-client-10.0.16-1.fc20.x86_64 (mariadb)
Requires: libc.so.6(GLIBC_2.14)(64bit)
 Error: Package: MariaDB-Galera-server-10.0.16-1.fc20.x86_64 (mariadb)
Requires: libc.so.6(GLIBC_2.17)(64bit)
 Error: Package: MariaDB-Galera-server-10.0.16-1.fc20.x86_64 (mariadb)
Requires: libc.so.6(GLIBC_2.15)(64bit)
 Error: Package: MariaDB-Galera-server-10.0.16-1.fc20.x86_64 (mariadb)
Requires: libc.so.6(GLIBC_2.14)(64bit)
 Error: Package: galera-25.3.5-1.sles12.x86_64 (mariadb)
Requires: libc.so.6(GLIBC_2.14)(64bit)
 Error: Package: galera-25.3.5-1.sles12.x86_64 (mariadb)
Requires: aaa_base
 Error: Package: galera-25.3.5-1.sles12.x86_64 (mariadb)
Requires: libstdc++.so.6(GLIBCXX_3.4.15)(64bit)

This doesn't look right. MariaDB fc20 are packages for Fedora Core 20.
Galera sles12 packages are for SUSE Linux Enterprise Server 12.

We have galera-25.3.5-1.sles12.x86_64 packages *only* in the sles12
repository, not in the centos6 repository.

We have fc20 packages only in the fedora20 repository, not in centos6
repository.

What did you mirror, exactly, and how?

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] doubt... show table status

[Sergei Golubchik]

Hi, Roberto!

On Dec 15, Roberto Spadim wrote:
 guys, i'm creating a table with:
 CREATE TABLE xxx SELECT * FROM 
 
 this one take a long time (more than 1 hour) to end...
 
 my problem is... i'm using HeidiSQL sometimes to check tables and etc...
 when i open database it execute a
 SHOW TABLE STATUS FROM `database`
 
 but is blocked while create table don't end, could we implement a non block
 show table? or a timeout?

yes, makes sense.
or, perhaps, SHOW TABLE STATUS should not show the table that is being
created?


Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Restated: JSON cannot represent binary data

[Sergei Golubchik]

Hi, Tom!

On Feb 03, Tom Worster wrote:
 Hi Sergei, 
 
 On 2/3/15, 7:13 AM, Sergei Golubchik s...@mariadb.org wrote:
 
  A dynamic column cannot be NULL, so using a JSON null (a different
  kind of null) to express dynamic column exists but cannot be
  represented as requested should work. The ORM would then have the
  names and positions in the structure of all the BINARY dynamic
  columns. With that it can send a SELECT with one or more
  COLUMN_GET(dyncol_blob, name AS BINARY) expressions. I could live
  with that.
 
 Hmm. May be a dynamic column cannot be NULL now, but this is not a
 conceptual limitation, there is no logical reason why it coldn't be.
 
 There may be no *technical* reason why it couldn't be but I see no
 *logical* reason why it would be. I can understand how tables (be they
 schema, temp or select) need NULL to say it's not here but a dynamic
 column existing and having a value of NULL is a non-sequitur. Writing a
 NULL to a dynamic column currently deletes it and I think that makes
 perfect sense. (I think we see know how the word column in dynamic
 columns can be misleading.)
...

Okay. I think that as our dyncols don't support NULLs and JSON does,
that might be a reasonable solution.

 I see what you mean but it's not a very useful answer. Much more
 useful to get the names, datatypes and structure plus all the values
 that aren't non-unicode strings.

Okay. Would you like to report a bug for that or should I?

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Serious problem with latest (10.0.16) Windows installer - destroys services.

[Sergei Golubchik]

Hi, Peter!

On Jan 31, Peter Laursen wrote:
 I have MariaDB 5.5, 10.0 and 10.1 installed (as Windows services named
 'maria55'. 'maria10' and 'maria10_1' ) and always update them to
 latest soon after an update is available. Yesterday morning I
 installed 10.0.16 on top of 10.0.15 using the .msi installer.  When I
 tried to connect to it a few hours later it was not possible.
...
 Questions:
 * Has somebody else experienced this?

Yes, https://mariadb.atlassian.net/browse/MDEV-7531

I've already fixed it and we've replaced msi packages with fixed ones.
Of course, it was already too late for you, I know :(

At least, I hope, it'll help those users who are not as fast as you are.

 * Has somebody replaced @wlad as maintainer of Windows installers?
 @wlad was the only person who ever was able to write reasonable
 MySQL/MariaDB installers. So I would regret that.

At the moment we don't have a dedicated Windows developer,
unfortunately.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] The future of Dynamic Columns

[Sergei Golubchik]

Hi, Tom!

On Jan 27, Tom Worster wrote:
  2. In at least two places (one was here
 https://mariadb.com/kb/en/mariadb/indexoptimizations-in-dynamic-columns/)
  I've read text with the implied meaning, we may improve such-and-such
 about MDC if and when people start actually using it.

While I know that there are users of dynamic columns, I don't remember
seeing any feature requests for them.

Okay, there was one, for JSON-like arrays, but it depends on having
array as a data type, like in regular columns. Not a standalone feature
for dynamic columns that we can just go no and implement.

   3. The MDC syntax is so clunky I worry that people who might find it
 useful are turned away.

That could as well be. And it's bad. What is worse - they're turned away
without complaining so we don't know what to do to fix dynamic columns.

Any suggestions how we can make the syntax less clunky?

   4. Did the 5.3 version with numbers for column names turn people off
   MDC for good?

As far as I have heard, those people who wanted dynamic columns in the
firts place found the implementation with numbers quite sufficient for
their needs. But new users probably didn't, I know :)

   5. The requirement to specify datatype when querying a dynamic column is
 agonizing. The manual says it is a limitation of SQL itself (presumably
 immutable) but PostgreSQL manages to avoid it when querying jsonb doc
 properties.

We could workaround this requirement by assuming everything is a string,
for example. This can be implemented, if that's what users want.

   6. The lack of indexing of dynamic columns is a severe limitation but
 PostgreSQL manages to index jsonb document properties.

Right, I know. We've heard this feature request a few times already.
That's why - no promises! - this feature is currently being discussed
for 10.2.

   7. I head a rumor that some kind of JSON features are coming to Maria.

I wonder what exactly you've heard and where :)
There are many different ways of implementing JSON features, some of
them actually use dynamic columns as the underlying storage format.

Currently there is no well defined JSON task in mariadb.org/jira,
so nobody knows what exactly will be implemented yet.

But now is the time to affect the 10.2 roadmap. What JSON features
would you like to see in MariaDB?

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Inclusion of Mysql security fixes in MariaDB

[Sergei Golubchik]

Hi, Raina!

On Jan 23, Raina Masand wrote:
 Hello,
 
 We recently were informed of some security fixes in Mysql 5.5.41:
 http://www.ubuntu.com/usn/usn-2480-1/ and are wondering whether there are
 plans to include these in an upcoming MariaDB release.  Right now, we are
 running 10.0.13, so we're trying to plan the next upgrade. We see that
 there have been similar fixes included in MariaDB 10.0.14 and 10.0.15, so
 this seems likely.
 
 Based on this https://mariadb.com/kb/en/mariadb/development/security/ list
 of CVE's, it looks like the MariaDB 10.0.15 and MariaDB 5.5.40 include the
 same security fixes (presumably pulled from Mysql 5.5.40). Can we expect
 that the fixes from Mysql 5.5.41 will be included in an upcoming MariaDB
 10.0.16 release? Would appreciate any insight into the general schedule for
 addressing these vulnerabilities.

Yes, I have updated the Security page to include these newly announced
vulnerabilities. They are fixed in MariaDB-5.5.41 and MariaDB-10.0.16.

Generally it works as follows:
* Oracle discovers or learns about a security vulnerability in MySQL
* Oracle doesn't tell anyone and secretly fixes it
* Oracle releases a new - fixed - MySQL version
* We (MariaDB) pull in MySQL changes and release a new MariaDB version
  - this usually takes few days (up to a week)
* Oracle releases a CPU with very vague description of vulnerabilities
  - http://www.oracle.com/technetwork/topics/security/cpujan2015-1972971.html
* By that time a fixed MariaDB version is already released, I only need
  to add new CVE numbers to the Security page

So, generally, when new vulnerabilities are publically announced,
the latest MariaDB release already has them fixed. Even if Security
page doesn't tell so.

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] build minimized MariaDB 10.0.x

[Sergei Golubchik]

Hi, Reindl!

On Jan 19, Reindl Harald wrote:
 https://mariadb.atlassian.net/browse/MDEV-7076
 
 well, in general i try to only have Aria/XtraDB/MyISAM
 
   -DWITH_XTRADB_STORAGE_ENGINE=ON \
   -DWITH_PARTITION_STORAGE_ENGINE=ON \
   -DWITH_ARCHIVE_STORAGE_ENGINE=OFF \
   -DWITH_BLACKHOLE_STORAGE_ENGINE=OFF \
   -DWITH_SPHINX_STORAGE_ENGINE=OFF \
   -DWITH_MROONGA_STORAGE_ENGINE=OFF \
   -DWITH_OQGRAPH_STORAGE_ENGINE=OFF \
   -DWITH_SPIDER_STORAGE_ENGINE=OFF \
   -DWITH_CASSANDRA_STORAGE_ENGINE=OFF \
 
 but OOM in context of mroonga
 
 i want my ./configure --help back while in other
 aspects cmake is really cool there is no way i know
 to get a list of valid configure options :-(

For cmake there is 'cmake -LH' and 'cmake -LAH' (the second also shows
advanced options).

But if a variable is never set in cmakefiles (which is the case for some
of WITH_xxx and WITHOUT_xxx - they are only checked, but never set), it
will not show out in help.

In 10.1 this is fixed in the sense that plugin configuration is now done
with PLUGIN_xxx variables. They show up in help, and they are sorted
separately, not intermixed with other WITH_xxx variables, like
WITH_EMBEDDED_SERVER or WITH_SAFEMALLOC.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Spider - query end

[Sergei Golubchik]

Hi, Nicolas!

On Dec 19, Nicolas Payart wrote:
 Hi list,
 
 I am using Spider Storage Engine for sharding data and remote access to 
 some innodb tables on 3 different MariaDB 10.0.14 instances.
 
 It performs quite well in production except that sometimes, some queries 
 (a minority of them fortunately) do not respond and stay indefinitely in 
 query end state in processlist.
 
 What can make a query stay indefinitely in query end state?
 Maybe a bug in Spider Storage Engine?
 
 Any advice is welcome!

I've asked the author of the Spider engine - Kentoku Shiba - and he
thinks that these queries don't stay indefinitely, but are just very
slow. They would be slow because, let me quote

  ... this case needs BKA join for better performance. But current table
  partition feature of official MariaDB does not support MRR well. In
  the other hand, there is a patched version of MariaDB that supports
  MRR well with partition feature. I will contribute this patch to
  official MariaDB.

meanwhile you can try the patched version of MariaDB (as prepared by
Kentoku) with BKA and see if that helps:

  Latest patched version of MariaDB source code
  
http://spiderformysql.com/downloads/spider-3.2/mariadb-10.0.13-spider-3.2-vp-1.1-mroonga-4.05h.tgz
  binary for Linux x86_64
  
http://spiderformysql.com/downloads/spider-3.2/mariadb-10.0.13-spider-3.2-vp-1.1-mroonga-4.05-linux-x86_64-glibc25h.tgz

Hmm, judging from the name this includes Spider 3.2, Vertical Partitioning
engine 1.1 and Mroonga 4.05h

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] the future of --skip-innodb

[Sergei Golubchik]

Hi, Federico!

On Dec 14, Federico Razzoli wrote:
 Hi.
 
 In MySQL 5.7 documentation:
 
 https://dev.mysql.com/doc/refman/5.7/en/innodb-parameters.html
 
 I read:
 
 As of MySQL 5.7.5, the InnoDB storage engine can no longer be
 disabled, and the --innodb=OFF and --skip-innodb options are
 deprecated and have no effect. Their use results in a warning. These
 options will be removed in a future MySQL release.
 
 Will MariaDB remove --skip-innodb? If so, when?

Currently we have no plans of removing or deprecating --skip-innodb.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] mariadb 10 build/link ignores specified linker path to specified lib (pcre). correct method?

[Sergei Golubchik]

Hi!

On Dec 10, grantksupp...@operamail.com wrote:
 building mariadb 10.0.15 on linux/64
 
 i have a non-system instance of (lib)pcre, installed in
 /usr/local/test/pcre, that i need to build/link against
... 
   cmake ..  \
...
-DCMAKE_C_FLAGS=... -I/usr/local/test/pcre/includes \
-DCMAKE_CXX_FLAGS=... -I/usr/local/test/pcre/includes \
-DCMAKE_SHARED_LINKER_FLAGS=... -L/usr/local/test/pcre/lib64 
 -Wl,-rpath,/usr/local/test/pcre/lib64 -lpcre \
...
 
 but id both cases, the resultant binary still links against system pcre libs
 
   ldd sql/mysqld | egrep -i pcre
   libpcre.so.1 = /usr/lib64/libpcre.so.1 (0x7f6165a5e000)
 
 what's the correct method for linking against a specified, non-system
 path'd lib?  specifically, here, libpcre?

Your command is almost correct.  You only need to set
CMAKE_EXE_LINKER_FLAGS, because CMAKE_SHARED_LINKER_FLAGS is used for
linking shared libraries.

Or you need to set both if you plan to use a shared library that needs
pcre (like ha_mroonga.so or libmysqld.so).

And you don't need to specify -lpcre, cmake adds it automatically and as
necessary.

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] mariadb 10 build/link ignores specified linker path to specified lib (pcre). correct method?

[Sergei Golubchik]

Hi, grantksupport!

On Dec 13, grantksupp...@operamail.com wrote:
 
 
 On Sat, Dec 13, 2014, at 12:45 PM, Sergei Golubchik wrote:
  Your command is almost correct.  You only need to set
  CMAKE_EXE_LINKER_FLAGS, because CMAKE_SHARED_LINKER_FLAGS is used for
  linking shared libraries.
  
  Or you need to set both if you plan to use a shared library that needs
  pcre (like ha_mroonga.so or libmysqld.so).
  
  And you don't need to specify -lpcre, cmake adds it automatically and as
  necessary.
 
 building with
 
   cmake ..  \
   ...
-DWITH_PCRE=auto \
-DBUILD_SHARED_LIBS=1 \
-DCMAKE_SKIP_BUILD_RPATH=0 \
-DCMAKE_BUILD_WITH_INSTALL_RPATH=0 \
-DCMAKE_INSTALL_RPATH_USE_LINK_PATH=1 \
-DCMAKE_INSTALL_RPATH=/usr/local/pcre/lib64 \
-DCMAKE_C_FLAGS=-I/usr/local/test/pcre/includes \
-DCMAKE_CXX_FLAGS=-I/usr/local/test/pcre/includes \
-DCMAKE_SHARED_LINKER_FLAGS=-L/usr/local/test/pcre/lib64 
 -Wl,-rpath,/usr/local/test/pcre/lib64 \
-DCMAKE_EXE_LINKER_FLAGS=-L/usr/local/test/pcre/lib64 
 -Wl,-rpath,/usr/local/test/pcre/lib64 \
   ...
   make
 
 still results in
 
   ldd sql/mysqld | egrep -i pcre
   libpcre.so.1 = /usr/lib64/libpcre.so.1 (0x7f5153b76000)
 
 something else missing? need changing?

Perhaps you're specifying all possible options and they affect each
other. I didn't try all that, I only used

cmake .. -DCMAKE_C_FLAGS=-I$HOME/pcre-local/include \
 -DCMAKE_CXX_FLAGS=-I$HOME/pcre-local/include \
 -DCMAKE_EXE_LINKER_FLAGS=-L$HOME/pcre-local/lib 
-Wl,-rpath,$HOME/pcre-local/lib

and that worked for me.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] R: Doubt about 'atomic' insert

[Sergei Golubchik]

Hi, Federico!

On Dec 13, Federico Razzoli wrote:
 INSERT SELECT is not standard SQL. But there is no reason to have a duplicate 
 value error.

Why do you think it's not? Here, in sql:2003 standard, part 2:

insert statement ::= INSERT INTO insertion target insert columns and 
source

insert columns and source ::= from subquery
  | from constructor
  | from default

from subquery ::= [ left paren insert column list right paren ]
[ override clause ]
query expression

query expression ::= [ with clause ] query expression body

and so on, from query expression body via query term to query primary
to simple table to query specification, which is

query specification ::= SELECT [ set quantifier ] select list table 
expression

So, INSERT INTO ... SELECT seems to be a valid standard syntax.

Regards,
Sergei

 
 Sab 13/12/14, Roberto Spadim robe...@spadim.com.br ha scritto:
 
  Oggetto: [Maria-discuss] Doubt about 'atomic' insert
  A: Maria Discuss maria-discuss@lists.launchpad.net
  Data: Sabato 13 dicembre 2014, 19:13
  
  Hi guysi'm
  with a doubt about the standard (sql standand?)
  error reporting , about INSERT SELECT...
  should this insert select return duplicate
  key?INSERT INTO errorsX (id) SELECT MAX(id)+1
  FROM errorsX;
  
  considering:CREATE
  TABLE errorsX( id INT NOT NULL DEFAULT
  0, PRIMARY KEY (id))
  ;


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Doubt about 'atomic' insert

[Sergei Golubchik]

Hi, Roberto!

On Dec 13, Roberto Spadim wrote:
 Hi guys
 i'm with a doubt about the standard (sql standand?) error reporting ,
 about INSERT SELECT...
 
 should this insert select return duplicate key?
 INSERT INTO errorsX (id) SELECT MAX(id)+1 FROM errorsX;
 
 considering:
 
 CREATE TABLE errorsX(
  id INT NOT NULL DEFAULT 0,
  PRIMARY KEY (id)
 ) ;
 

Depends on how many rows the table errorsX has.

If it has only one row, than this INSERT ... SELECT will insert a one
row with a different id. No error.

If it has more than one row, this INSERT ... SELECT will try to insert
more than one row, all with the same id value - there will be a
duplicate key error.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.15 now available

[Sergei Golubchik]

Hi, Peter!

On Dec 09, Peter Laursen wrote:
 
 But most important TokuDB plugin does not load. This may because this is
 SuSE 12.3 (that may ship with another  C-runtime etc. than what 12.1 and
 12.2 did).
 INSTALL SONAME 'ha_tokudb';
 -- Error Code: 1123
 -- Can't initialize function 'TokuDB'; Plugin initialization function
 failed.

See the error log, it might contain more specific error messages.
One probable reason could be transparent_hugepage enabled on your
computer.

 BTW: I also had some communication on Fedora's mailinglist. They have
 MariaDB 10.0.14 in Fedora 21 repos.  But without TokuDB.  They cannot build
 it. They have reported to MariaDB (they say).

Yes, there were issues with TokuDB and gcc 4.9 (MDEV-6446, MDEV-6449),
but they were fixed.

And there was binutils bug in Fedora (it resulted in crashes when
compiling TokuDB): https://bugzilla.redhat.com/show_bug.cgi?id=1149660
As far as I understand, it's also fixed by now.

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.15 now available

[Sergei Golubchik]

Hi, Daniel!

Perhaps we should put a note for every new platform that we add (was
CentOS 7, now it's openSUSE) that this is a new platform, just added,
a first release for it, so while binaries are same as everywhere and
stable, the packaging (conflicts, dependencies) is new and might have
some rough edges?

Peter, 
I've created MDEV-7211 for this issue to make sure it won't be
forgotten.

Regards,
Sergei

On Nov 25, Peter Laursen wrote:
 I *CERTAINLY* think this
 https://downloads.mariadb.org/mariadb/repositories/#mirror=23Mediaversion=10.0distro_release=opensuse13-amd64distro=openSUSE
 is
 premature taken this discussion into account!
 
 And I believe this You will also be prompted about a conflict between
 openSUSE's own MariaDB 5.5 client package and the MariaDB client package
 is wrong. It may be correct for 12.2 but 12.3 does not have the MariaDB 5.5
 client package. It had 10.0.13 (and I think also some client package from
 Oracle - probably for some compability reason.
 
 What about the conflicts on OpenSuSE 12.3 *1) with KDE and *2) with MariaDB
 10.0.13 as installed from the distro DVD?
 
 -- Peter

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.15 now available

[Sergei Golubchik]

Hi, Peter!

On Nov 26, Peter Laursen wrote:
 
 (and please add TokuDB and InfiniDB soon. ¨'playing around' with TokuDB was
 my primary motivation for all this)

As far as I can see, tokudb is present in
MariaDB-10.0.15-opensuse13_1-x86_64-server.rpm.

Did you install the plugin?

INSTALL SONAME 'ha_tokudb'

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.15 now available

[Sergei Golubchik]

Hi, Peter!

On Nov 26, Peter Laursen wrote:
 NO.. I thought it would bea separate RPM (Like the OQGraph). I will
 check when I get time.  it coudl take a day or two now. Sorry for
 false alarm.

As a general rule we put in separage packages plugins that require
additional dependencies. OQGraph requires libJudy, so if one doesn't use
OQGraph, she shouldn't be required to install libJudy. CONNECT needs
odbc, xml, etc. TokuDB doesn't bring in extra dependencies, so it's
included into the server package.

 ButSuSE should  be listed here
 https://mariadb.com/kb/en/mariadb/documentation/storage-engines/tokudb/how-to-enable-tokudb-in-mariadb/
 then, right?

Added, thanks.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.15 now available

[Sergei Golubchik]

Hi, Peter!

On Nov 26, Peter Laursen wrote:
 If TokuDB has no dependencies, that the server does not have, why is it
 then not available on Windows? If it is, then Windows should be listed on
 same page.

It doesn't compile on Windows, as far as I understand.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB 10.0.15 now available

[Sergei Golubchik]

Hi, Peter!

On Nov 26, Peter Laursen wrote:
 OK .. but did the TokuDB guys ever try to figure out why?
 
 It could be a good idea to do. Because if it fails with Visual Studio, it
 may fail with other compilers as well. GCC is only one compiler and not the
 most standards-compliant as far as I understand.

This is something that TokuDB developers need to do. They apparently
don't build on Windows yet. If upstream will support Windows, we will
most certainly include TokuDB in our Windows packages.

May be if you'd ask them... :)

Regards,
Sergei

 -- Peter
 
 On Wed, Nov 26, 2014 at 4:02 PM, Sergei Golubchik s...@mariadb.org wrote:
  On Nov 26, Peter Laursen wrote:
   If TokuDB has no dependencies, that the server does not have, why
   is it then not available on Windows? If it is, then Windows should
   be listed on same page.
 
  It doesn't compile on Windows, as far as I understand.
 

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB for OpenSuSE 12.3

[Sergei Golubchik]

Hi, Peter!

On Nov 24, Peter Laursen wrote:
 Thanks Elena. That even I can understand.
 
 But error on installation of server packet No packet found allowing for
 the action specified. I am simply launching YaST graphical interface from
 the context menu of the RPM file placed on my desktop (install with packet
 manager). There is a reference to some log with more details, but I don't
 know where it can be found.

I've tried to repeat this - but over ssh, so I used text-mode yast2.
Supposedly, selecting install with packet manager from the context
menu is equivalent to sudo yast2 -i /path/to/file.rpm (I don't know
for sure).

Here's the issue. MariaDB-server package depends on other MariaDB-*
packages. Normally yast or yum fetch dependencies automatically from a
repository. But you don't install from a repository, you install a
specific file. In this case yast cannot fetch dependencies. You have to
install either all packages in one command, like this:

  sudo yast2 -i MariaDB*server.rpm MariaDB*client.rpm MariaDB*common.rpm

or you can do it with GUI and a context menu - but you need to install
MariaDB*common.rpm first, then MariaDB*client.rpm, and only then you can
install MariaDB*server.rpm.

I tried both solutions - both worked.

Regards,
Sergei

P.S. Now, when we build for SuSE, I hope we will provide a proper
package repository and yast will be able to resolve dependencies automatically.


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] MariaDB for OpenSuSE 12.3

[Sergei Golubchik]

Hi, Peter!

We've recently started building on SuSE:
http://buildbot.askmonty.org/buildbot/builders/kvm-zyp-opensuse13_1-x86_64
http://buildbot.askmonty.org/buildbot/builders/kvm-zyp-opensuse13_1-x86

This is a new builder and we haven't made any releases of these
binaries yet, as far as I remember.

Still the source code is the same, so it is as stable as any other 10.0
build. But the packaging - like dependencies or conflicts - might be not
completely polished yet.

So, if you'd like, you can try our suse rpms that were built on opensuse,
as above.

On Nov 23, Peter Laursen wrote:
 I have OpenSuSE 12.3.  It ships with MariaDB 1.0.13 - and with no TokuDB
 and no Galera options. I would like to upgrade to 10.0.14 and also enable
 TokuDB. And it does not seem that an upgrade will be available from SuSE
 software repositories.
 
 There are a lot of RPMs in the yum repository
 http://mirror.23media.de/mariadb/mariadb-10.0.14/yum/. But does anyone know
 if they will work with SuSE 12.3? Runtime environment (kernel, glibc etc.)
 should be compatible of course, and also  RPMs for Redhat/Fedora/CentOS
 systems will not alwyas work on SuSE, as SuSE has some specific
 requirements for the SPECS.
 
 When I used SuSE 10.x many years ago, the generic glibc23 RPMs available
 worked perfectly.
 
 Note: I *only* want to install a server that can be handled by SuSE's YaST
 package manager! So suggestons for any other solution (such as using the
 tarball) is not an option. Or it would be last resort.
 
 I would not mind give it a try if someone can advise what I should try. I
 have SuSE running in a VM and I will be able to return to a snapshot of the
 system easily if something goes wrong.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Excluding plugins when building MariaDB 10.1

[Sergei Golubchik]

Hi, Kolbe!

On Sep 03, Kolbe Kegel wrote:
 Hi!
 
 I'm trying to build MariaDB 10.1 on OS X.
 
 cmake ../server/ -DBUILD_CONFIG=mysql_release -DWITH_SSL=system 
 -DCMAKE_CXX_FLAGS:STRING='-stdlib=libstdc++' -DPLUGIN_TOKUDB=NO 
 -DPLUGIN_OQGRAPH=NO -DWITH_EMBEDDED_SERVER:BOOL=OFF -DWITH_WSREP=OFF
 
 The build is failing on tokudb code even though I am trying to avoid building 
 the TokuDB plugin. What gives?
 
 Linking CXX shared library libft.dylib
 Undefined symbols for architecture x86_64:
   std::__1::__vector_base_commontrue::__throw_length_error() const, 
 referenced from:
   toku::dmtklpair_struct, klpair_struct*, 
 toku::klpair_dmtwriter::verify() const in bndata.cc.o
   std::__1::__vector_base_commontrue::__throw_out_of_range() const, 
 referenced from:
   toku::dmtklpair_struct, klpair_struct*, 
 toku::klpair_dmtwriter::verify() const in bndata.cc.o
   toku::dmtklpair_struct, klpair_struct*, 
 toku::klpair_dmtwriter::verify_internal(toku::dmt_internal::subtree const, 
 std::__1::vectorbool, std::__1::allocatorbool *) const in bndata.cc.o
 ld: symbol(s) not found for architecture x86_64
 clang: error: linker command failed with exit code 1 (use -v to see 
 invocation)
 make[2]: *** [storage/tokudb/ft-index/ft/libft.dylib] Error 1
 make[1]: *** [storage/tokudb/ft-index/ft/CMakeFiles/ft.dir/all] Error 2
 make: *** [all] Error 2

What MariaDB version do you have?

I've just tried this exactly command line on OSX on 10.1 - worked as
expected, TokuDB was not built.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Excluding plugins when building MariaDB 10.1

[Sergei Golubchik]

Hi, Kolbe!

On Sep 04, Kolbe Kegel wrote:
 On Sep 3, 2014, at 11:20 PM, Sergei Golubchik s...@mariadb.org wrote:
  On Sep 03, Kolbe Kegel wrote:
  
  I'm trying to build MariaDB 10.1 on OS X.
  
  cmake ../server/ -DBUILD_CONFIG=mysql_release -DWITH_SSL=system 
  -DCMAKE_CXX_FLAGS:STRING='-stdlib=libstdc++' -DPLUGIN_TOKUDB=NO 
  -DPLUGIN_OQGRAPH=NO -DWITH_EMBEDDED_SERVER:BOOL=OFF -DWITH_WSREP=OFF
  
  The build is failing on tokudb code even though I am trying to
  avoid building the TokuDB plugin. What gives?
  
  What MariaDB version do you have?
 
 I was building a very recent github pull of MariaDB 10.1.
 
 commit d161546b67142cdd5322a4ed160441045ae0cd1e
 Author: Sergei Petrunia pser...@askmonty.org
 Date:   Thu Sep 4 01:12:49 2014 +0400
 
  I've just tried this exactly command line on OSX on 10.1 - worked as
  expected, TokuDB was not built.
 
 How strange...
 
 I used -DWITHOUT_TOKUDB=1 as suggested by Kristian, and that built OK,
 but -DPLUGIN_TOKUDB=NO doesn't seem to be doing the trick.

Yes, old WITH_xxx/WITHOUT_xxx variables are still supported.
They're internally converted to PLUGIN_xxx values.
But PLUGIN_xxx should work on their own too.

 I have cmake 3.0.1, in case that's relevant.

Yes, that's it. I only tried with 2.8.12.2.
Now, with 3.0.1 I can repeat this too, thanks.

Regards,
Sergei

___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp

Re: [Maria-discuss] Upgrade from 10.0.12 - 10.0.13 changes default config file location behavior; build-time -DDEFAULT_SYSCONFDIR= is now ignored ?

[Sergei Golubchik]

Hi, grantksupport!

On Aug 12, grantksupp...@operamail.com wrote:
 I build MariaDB from src.
 
 I've been running v10.0.12 (r4252).  I've upgraded to v10.0.13 (r4346).
 
 My build config always includes
 
   cmake ..  --debug-output \
   -DCMAKE_BUILD_TYPE=Release \
   -DCMAKE_INSTALL_PREFIX=/usr/local/mariadb \
   -DINSTALL_LAYOUT=STANDALONE \
   -DINSTALL_SYSCONFDIR=/usr/local/etc/mariadb.DEFAULT \
   -DINSTALL_SYSCONF2DIR=/usr/local/etc/mariadb.DEFAULT/conf.d \
   -DDEFAULT_SYSCONFDIR=/usr/local/etc/mariadb \
   ...
 
 I have always kept my config only in
 
   /usr/local/etc/mariadb/my.cnf
 
 with no ~/.my.cnf in existence.
 
 Launching MariaDB (via systemd) with
 
   /usr/local/mariadb/bin/mysqld 
 --defaults-file=/usr/local/etc/mariadb/my.cnf --basedir=/usr/local/mariadb 
 --datadir=/var/db/mariadb --plugin-dir=/usr/local/mariadb/lib/plugin 
 --log-error=/var/log/mariadb/mariadb-err.log 
 --pid-file=/var/cache/mariadb/mariadb.pid 
 --socket=/var/cache/mariadb/mariadb.sock --port=3306
 
 with MariaDB v = 10.0.12, changes to my config have always been
 correctly picked up, INCLUDING by all mysql client tools.
 
 So, for example, if in my.cnf I add
 
   ...
   [client]
   user = root
   password = 'XX'
   ...
 
 then at shell, instead of
 
   mysqlshow -u root -p
   password: XX
 
 I can just
 
   mysqlshow
 
 and the options as specified are picked up.

I don't see how this could've been possible. The code doesn't search in
the @CMAKE_INSTALL_PREFIX@/etc, and I've just built MariaDB 10.0.12, and
tools aren't reading /usr/local/mariadb/etc/my.cnf file:

$ mysqldump --help
...
Default options are read from the following files in the given order:
/etc/my.cnf /etc/mysql/my.cnf /usr/local/etc/mariadb.DEFAULT/my.cnf ~/.my.cnf 
...

And 10.0.13 didn't change that.

Perhaps you've symlinked  /usr/local/mariadb/etc/my.cnf to /etc/my.cnf?
Because this was changed in 10.0.13, indeed.

Before 10.0.13 MariaDB was reading /etc/my.cnf and /etc/mysql/my.cnf
unconditionally. Since 10.0.13 it does that only if you didn't specify
INSTALL_SYSCONFDIR or DEFAULT_SYSCONFDIR. If you've explicitly specified
where your config files are, MariaDB won't look elsewhere, you can
sandbox it in a dedicated location and know that it won't look outside
of it.

Regards,
Sergei


___
Mailing list: https://launchpad.net/~maria-discuss
Post to : maria-discuss@lists.launchpad.net
Unsubscribe : https://launchpad.net/~maria-discuss
More help   : https://help.launchpad.net/ListHelp