[peterlaursen]

You shall not think in terms of files. Only the MySQL server can use those files directly. To access the data in the files you must use a MySQL client like SQLyog. KEEP YOUR HANDS OFF those files. MySQL does what needs to be done!

When you connect to the running MySQL server dont the database name show up in the left column? Now expand it using the + icon and tables will be shown. Click a table and click the DATA tab and your data wil be shown.

BTW: start reading the SQLyog help file and next the FAQ …

[peterlaursen]

It is a pretty old verison you are using, I guess?

Where did you get if from?

Could I ask you to try 5.1 beta (trial also available here).

This new version should fix issues with enocdings, as it issues a 'SET NAMES = uft8' on both servers.

Are you using HTTP-tunnelling?

If yes there may still be a problem (to be fixed in BETA 7), but I think we can fix it with hard-coding the tunnelling file …

[peterlaursen]

We will need more info to be able to answer this. To answer we must first understand and to understand we must reproduce. 😛

Is it possible that you attached a – not too big – test case to reproduce.

Also read: http://www.webyog.com/faq/1_26_en.html

[peterlaursen]

@ritesh – regarding 3)

'at the end of a word' just means 'if there is no letter after cursor position' or 'only if there is a , a TAB, a LINEBREAK or NOTHING after cursor position'.

That is for instance 'her' triggers autocomplete end

'here' does not.

Autocomplete should append to strings only .. not be functional in the midst of a string.

[peterlaursen]

You mean beta 6, don't you?

Quote:

… but it would be good to get some more help on this.

Well it was not before Thursday an Friday that information was coming up, thatcould be a 'key' to this. Now it has been weekend. But now start a new working week ..

Your observation about the 'favorites' folder being related to this, might very well be the beginning of the solution.

What is your OS and the file system?

[peterlaursen]

Hmmm .. this is 'hairy' …

There is something crazy with your data. I can import your MYSQLDUMP data. I can export with SQLyog, but not import these files again as there are 'empty insert statements' in the file.

However if you empty or delete the three tables where sync fails at target, the sync runs OK. The data are created at the target. This seems permanently to solve the issue with the two last tables (I can edit and add data and sync)

But not the first table – the source_objectcache table. There is something very very weird with the source_objectcache.value column. I cannot update the 'exptime' value or delete a row in this table from GUI because of this! However an update staement like

Code:

update source_objectcache set exptime = '2006-04-09 19:12:40' where keyname = 'SOURCE:messages';

works fine. But not if the .value column is included in the where!

Looks like the SQLyog/SJA code has an issue with this MEDIUMBLOB column (it cannot be because of the binary collation!!!)! The SYNC tool does not identify them correctly and even a simpel WHERE clause in an update statement with these data fail.

But for the last two tables I simply think your target database (indexes) was corrupt. If I delete those tables at target and let the sync recreate them, there is no issue anymore no matter what I do at source. ('Repair Tables' is not an option with InnoDB tables)

Over to Bangalore 😀

@ritesh: BTW: why does not the WHERE simply use the PK as it did before? I think that change came with the 'defaults' thing of 5.1. But imagine a LONGBLOB of 4 GB in a WHERE …

Your data attatched as a ZIP package for convenience

[peterlaursen]

Iam sorry …

Seems that I sync'ed the same file to itself.

With the correct data for source and target i get these errror:

Code:

Sync started at Sun Apr 09 23:13:22 2006

Error No. 1062
Duplicate entry 'SOURCE:pcache:idhash:1-0!1!0!0!!en!2' for key 'keyname'

Error No. 1062
Duplicate entry '1' for key 'si_page'

Error No. 1062
Duplicate entry '1' for key 'ss_row_id'

Now I see what you mean. Three rows fails to sync. They are:

source_objectcache: 3rd row causes trouble //note latin1_bin collation (however collations should not matter!).

source_searchindex: 1st and only row causes trouble //note UNSIGNED integer.

source_site_stats: 1st and only row causes trouble //note UNSIGNED integers.

Yes there is an issue here! Those three rows should have been UPDATED. Instead SJA tries to INSERT, but the server prevents this as the PK's are identical. And that is what the error message is about.

I am in doubt whether your database(s) is/are corrupted or if it is an issue with SJA/SQLyog.

I experience that when SQLyog exports the data the export cannot import!

[peterlaursen]

Hi again ..

Your data sync'ed fine here between two MySQL 5.0.20 databases with SJA for Windows as of SQLyog 5.1 beta6 on this local Windows machine.

Code:

SQLyog Job Agent Version 5.1
Copyright (c) Webyog Softworks Pvt. Ltd.. All Rights Reserved.

Sync started at Sun Apr 09 22:23:42 2006

Table SrcRows TgtRows Inserted Updated Deleted
========================= ======= ======= ======== ======= =======
`source_archive` 0 0 0 0 0
`source_categorylinks` 0 0 0 0 0
`source_hitcounter` 0 0 0 0 0
`source_image` 0 0 0 0 0
`source_imagelinks` 0 0 0 0 0
`source_interwiki` 174 174 0 0 0
`source_ipblocks` 0 0 0 0 0
`source_logging` 0 0 0 0 0
`source_math` 0 0 0 0 0
`source_objectcache` 3 3 0 0 0
`source_oldimage` 0 0 0 0 0
`source_page` 0 0 0 0 0
`source_pagelinks` 0 0 0 0 0
`source_querycache` 0 0 0 0 0
`source_recentchanges` 2 2 0 0 0
`source_revision` 1283 1283 0 0 0
`source_searchindex` 1 1 0 0 0
`source_site_stats` 1 1 0 0 0
`source_text` 1283 1283 0 0 0
`source_trackbacks` 0 0 0 0 0
`source_transcache` 0 0 0 0 0
`source_user` 1 1 0 0 0
`source_user_groups` 2 2 0 0 0
`source_user_newtalk` 0 0 0 0 0
`source_validate` 0 0 0 0 0
`source_watchlist` 0 0 0 0 0

Total time taken – 11 sec(s)

So it could be an issue that has been fixed lately (there have been a lot of fixes). Porbably SJA for Linux will be recompiled too when SQLyog 5.1 is FINAL. Do you have a chance to install SQLyog on WINE? The SJA for windows will run with WINE. Refer to: http://www.webyog.com/faq/31_71_en.html .

I have to leave for an hour but will test later with SJA for Linux version 5.02 (that is now 4-5 months old – yours is even older). I will copy data to my SuSE10 that also runs 5.0.20. Check back in 1-2 hours!

I think that there are UTF-8 data in the file. That probably is where 'the chain breaks' with your SJA version

To start it the command is just

Code:

wine sja jobfile.xml

?

BTW: to attach data here you should ZIP then. We support a few graphics formats directly – else only ZIP! We shalll not risk any sort of 'code injection' or 'SQL injection' to our server<_<

[peterlaursen]

I cannot reporduce anything like this (just synced a lcoal and remote MySQL 4.0.26 successfully with HTTP-connection to the remote host). I have the option to try a 4.1 and 5.0 server as well, but I'll need more info about what to test!

There must some specific reason for this! Please refer to http://www.webyog.com/faq/1_26_en.html for details that are always must appreciated when reporting an issue.

Error No.1046 No database selected could be a charset issue. Are there any special characters in the database name?

You may post a simple reproduceable test case here. Preferably two very small dumps (a few rows of data will do – as long as the problem persists with them) of two DB's that behave like this

[peterlaursen]

Well I tried creating this DB:

Code:

create database if not exists `mystest`;

USE `mystest`;

/*Table structure for table `myt` */

DROP TABLE IF EXISTS `myt`;

CREATE TABLE `myt` (
`id` bigint(20) NOT NULL auto_increment,
`ttt.ttt` varchar(50) default NULL,
PRIMARY KEY (`id`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1;

/*Table structure for table `tttttt` */

DROP TABLE IF EXISTS `tttttt`;

CREATE TABLE `tttttt` (
`id` bigint(20) NOT NULL auto_increment,
`ttt.ttt.ttt` varchar(50) default NULL,
PRIMARY KEY (`id`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1;

And things work 100%. Also columns with a “.” is now found by autocomplete OK.

Looks like indexes were currupted in the old file …

But backticks should still be stripped!

[peterlaursen]

This crash is reproducable here .. even when I don't have the TAGs file of yours.

To reproduce

1) copy the complete view definition to editor

2) delete `msgs`.`quar_type` AS `Quarantine`,

3) type `msgs.

and it crashes.

If I only type the `msgs. into an empty editor there is no crash

Let me point here to two more issues with autocomplete.

(1 is more important than 2. Actually I think 1) is a blunder 😀 )

1)

select * from ta finds table_name

but

select * from `ta does not.

I think autocomplete should strip backticks before evaluating

2)

select * from table_name where co

finds “column_name” but not “column.name.“

I think punctuations are valid inside a columnname ??

[peterlaursen]

1)

Quote:

Ideally, (IMO) auto-complete should only be offered at the end of a word, or after a .

By design auto-complete reads from last to cursor position and appends there. It does not intend to read what (or if) is after the cursors. Well .. I think you should only use autocomplete where it make sense to use it.

2)

If there are more alternatives possible it switches between them as you press TAB again. When first BETA was out I posted this: http://www.webyog.com/forums/index.php?showtopic=1944 . That is how it is supposed to work.

[peterlaursen]

Quote:

I have also tried to contact webyog without any results.

How did you try this? If you mail at supportATwebyogDOTcom you will have an automated reply with a link to the ticket system (we have had to close this mail adress due to ever-increasing amounts of spam and virus-mails). Maybe this automated mail was lost too?

I can't find any 'ticket' in our system relating to your name, and there is no such request that has not been answered (checked it 2 minutes ago!).

Create one here: http://www.webyog.com/support/ttx.cgi

.. and don't forget to tell what email adress you used for the upgrade-request. We will then easily be able to find a copy of the mail you should have had. We'll then post details for you there.

Also write down the ACCESS Key and the ticket no. You can access the ticket from the adress

http://www.webyog.com/support/ttx.cgi?cmd=ticket&sid=xyz ..where xyz is the number

… just in case that the mail with these details also gets lost!

We are working on this issue with mails that don't arrive. Admittedly we have had quite a few over the last weeks. They are sent from the Webyog SMTP-server as the cc-copy for 'us-ourselves' arrive as it should! It has been a mystery how they get lost. Looks like the ISP hosting our server has some problems with sending mails to certain domains. Maybe Ritesh has some fresh information …

It is very annoying for us too, and we and the ISP do all we can to solve this thing.

You did check all spamfilters you have (on you system, at your ISP) ?

[peterlaursen]

Quote:

In those tables that do not correctly synchronize there are no rows with identical rows

No but the MySQL server claims that there are identical Primary Keys. The error 1062 is a MySQL server error – not an internal SQLyog error. Also the Selects don't tell much as any insert of data with a dublicate key would fail. There are no dublicate keys in the select that you post here, but the Server would never allow for creation of dublicate primary keys!

To understand how SQLyog DATA SYNC uses the Primary Key, read http://www.webyog.com/articles/Using_SQLyo…L_Databases.pdf

Is it possible that you could provide a 'test case'. 'create staements for the tables' (rather than 'describes')and two dumps (source and target data – not too many) to reproduce the issue

BTW: what SQLyog program verison are you using?

[peterlaursen]

What I say is that a sequence of the ASCII control characters (or 'rn' in C-syntax notation or or (in HEX) “0D 0A”) is stripped out somehow, when you load an existing XML file into the wizard.

And as there are not such characters the Notifications SQL Window does not display a linebreak. It displays what data it has!

Try: Run the wizard 'from scratch', write a statement of two lines and copy to a HEX-editor (mark-up with the mouse and use CTRL-C keyboard shortcut to copy to clipboard, netx past (CTRL-V) into HEX-editor). Finish the wizard and save the XML. Next run the wizard again, open the exiting XML, step through the wizard till Notifications SQL window appears and copy again to HEX-editor the same way. Compare the two HEX views. What you copied in the first run has the “0D 0A” HEX sequence (that is a WINDOWS linebreak 'code') – what you copied in the second run has not.

[Author]

Posts