Forum Replies Created
-
AuthorPosts
-
peterlaursen
ParticipantWe will follow up with our own example populated with data ASAP.
Yesterday we forgot to turn off Power management, so it did not finish! .. 🙁
peterlaursen
ParticipantOn my laptop (a “Athlon Turion64x2 mobile technology” (2*1.8 ghz)), 1 G RAM with Vista 32 bit
.. it takes (without data) ~55 seconds.
Server: 5.0.51b
InnoDB configuration parameters (as set up by the Windows installer/configuration wizard):
innodb_additional_mem_pool_size=2M
innodb_flush_log_at_trx_commit=1
innodb_log_buffer_size=1M
innodb_buffer_pool_size=18M
innodb_log_file_size=10M
innodb_thread_concurrency=8
I think that either of those
* InnoDB configuration parameters
* A MySQL/InnoDB issue with your CPU architecture.
.. is the culprit here
Also I do not believe 'Xeon' is an exact description. 'Xeon' is an Intel tradename/brand for high-level (fast and with large cache) x86 processors only. There are different 'Xeon' processors. I don't know full details however – only this:
http://en.wikipedia.org/wiki/Xeon . “The Xeon brand has been maintained over several generations of x86 and x86-64 processors”
I wonder if it is 64 bit and the OS and the MySQL server is 64 bit too?
If you can reproduce significant difference in performance with SHOW FULL FIELDS (and other commands) on this hardware compared to a 'normal PC' with same InnoDB settings, I think you should report to bugs.mysql.com.
peterlaursen
ParticipantIt is still not implemented.
We do all we can with the resources available.
peterlaursen
Participant1) The logic of the Structure Sync code will make it impossible for us not to consider column order. Also to prepare the databae for Data Sync the column order must be the same. This we won't change.
2) We could probably provide an option to ignore collation differences, but if you “don't care about it” why is it so important?. You should be aware that concatting, joining, using subqueires (and much more) data with different collations is not possible. For 'non-trivial queries' collations are important! I do not think we will do this either, but this we will consider again.
peterlaursen
ParticipantWe have released 6.55RC.
This is not a complete fix for this. You may still see a few pixels missing in the right side. But not as much as before!
peterlaursen
Participantzips are damaged or empty!
(tried different zippers and tried direct FTP download from the server).
Please try here instead:
peterlaursen
Participant“Is it possible that the slowness is due to the fact theres 30 or so of these databases in the browser ?”
I doubt this! SHOW FULL FIELDS will only care about the database on which it is executed.
I suspect that your InnoDB configuration settings are *much too agressive* or that MySQL and/or does not scale well on this CPU architecture. Or some issue with this specific server version.
The test we did wiht empty tables was wiht a *tiny* configuration (a 'developer machine' setting as generated by the configuration wizard bundled wiht the Windoes installer).
peterlaursen
ParticipantWe create a database with 900 tables with each 15-20 columns. Empty tables sync in 1:05 minute (on an 'ordinary' PC – Core Duo 2*2 Ghz).
We will populate with data tonight and see the difference.
.. but we really would appreciate you table structure.
Also the slowness in Object Browser is not reproducable for us (but also here we do SHOW FULL FIELDS).
We cannot explain why this is so slow on your machine.
table structure .. PLEASE …
peterlaursen
ParticipantAnd still: if you could provide a structure-only dimp it would be very nice.
peterlaursen
ParticipantHello ..
It is important for us to know if structurally identical but empty tables are equally slow or not.
With SQLyog you can easily create such tables with the 'duplicate table' functionality.
You may also use 'copy to other' and only copy structure for the two databases involved and will have two databases that are structurally identical (but empty) as compared to the original ones (to use 'copy to other' you need two connections to the same server only).
peterlaursen
ParticipantThe file is php.ini. But on 'shared hsoting' it is unlikely you can change for yorself beecause as I wrote the file is shared between all users on the host and the webserver will need to be restarted after the change. So some server administrator will need to do this.
If you upload a small PHP file only calling the 'phpinfo()' function you can see the current values
peterlaursen
ParticipantI do not have your tables structures etc. but if I make everything comments
Code:DELIMITER $$DROP TRIGGER /*!50032 IF EXISTS */ `metalsmith`.`t_inventoryAI` $$
CREATE
TRIGGER `test`.`t_inventoryAI` AFTER INSERT ON `test`.`abstracts`
FOR EACH ROW BEGIN
DECLARE currentAmt decimal(17,10);/* SELECT SUM(current_amount) INTO currentAmt
FROM inventory i
WHERE i.product_id = NEW.product_id;*/
— one comment
# UPDATE product p
— SET p.current_amount = currentAmt – p.reserved_amount,
#one more comment
— p.cost = f_getAvgCost(NEW.product_id)
#WHERE p.product_id=NEW.product_id;
/* some more comment */
END;
$$DELIMITER;
'execute all' and from SQLyog select 'alter trigger' for that trigger it displays:
Code:DELIMITER $$DROP TRIGGER /*!50032 IF EXISTS */ `test`.`t_inventoryAI`$$
CREATE
;I can isolate one of the INSERT lines that is taking characters and messing them up if needed.
Thanks,
CALICO
peterlaursen
ParticipantWith Stored Procedures, Functions, Views and Events what we are displaying is what 'SHOW CREATE …' returns. So in these cases it is the server that does the formatting.
With Triggers we build the CREATE statement generically (by querying Information_Schema) as there is no 'SHOW CREATE TRIGGER' statement.
In particular with VIEWs the server totally rewrites the statement at create time. This is because as per SQL standards a VIEW is 'defined frozen' at create time. If you
* create view aview as select * from atable
* alter atable add column ..
… then the view still references ONLY the columns existent at create time. So far it is reasonable and correct. But the formatting of a View as returned by the server is horrible!
What is your server version?
Can you post a few examples of this? But probably you should first ensure to have a recent server version!
peterlaursen
ParticipantI think you cannot by yourself! The PHP configuration is probably in a shared php.ini file.
Maybe you can ask the Sysadmin to raise MAX_POST setting in PHP configuration. Default is (most often) 8 MB.
I am not sure this is the issue, but worth a try!
peterlaursen
ParticipantDoes the error happen in the middle of a session on those sites or only when connecting for the first time?
It could bee some PHP buffer setting that is not large enough for large result sets. Such applications may have some big BLOBs. But that is only a guess.
-
AuthorPosts