RE: [firebird-support] Firebird replication help needed

[drci...@yahoo.com [firebird-support]]

Hello Alan, Thank you for your answer. I will download the FB Replicator try it 
and see if it will work for me. 
 That part with the triggers i've already done. I have after I, U, D triggers 
for every table and they insert into my replication log table. The problem is 
that I used to put the sql query that was executedon the client DB on this 
table in the log and then just execute it on the server DB. But if the table 
metadata is changed, then that doesn't work anymore, cos its says can't execute 
to server because the fields in the insert do not equal to those in the table. 
So I decided insted to log only the table name, the PK of record, timestamp and 
the type of query - I U D. And if it's I, to execute insert or update on 
server. But probably when I execute them I'll have to separate them by table 
and get every field name and then value of that table and then onstruct the 
query as a string so it will be probably slow. 
 My other problem is should I do it in the C# replication app of use a script 
instead?
 Thank you again.If I have further questions I'll write.

RE: [firebird-support] Firebird replication help needed

['Alan McDonald' a...@meta.com.au [firebird-support]]

I am writing a two way server - client Firebird replication with a C# app and I 
am facing the following problem. I have a log table with the table name pk of 
the record and type of the query executet on it. For example if in the sales 
table there has been inserted a line with primery key of 4 there will be the 
table name, pk of the record and insert. My idea is if the statement is insert 
to take every value from the client's DB and insert it into the server's. How 
do I do this and how to do it so that it can work fast. Should I devide every 
record by tables and select all unreplicated records from that table or there 
is some other way. Please help me.

By the way I used the method where every query that has been executed is 
written in a fild in the log table, but that makes it impossible to add columns 
to the table, cos the query will be wrong. Please help me. I've been strugling 
to write a good replication for years.

 

If you want, you can have the source for FBReplicator.

Replication Manager, Service and Service Controller.

The manager allows for you to update table structures and install and maintain 
triggers in the source DBs such that all I, D and U operations are stored in a 
changes table. The service then runs at your desired interval and replicates to 
a target. You can have 2-way replication.

I’ve been running 3 way replication (i.e. 4 servers each puching 3 ways) since 
1997. That’s 72 server years of 24x7 traffic.

 

Alan

[firebird-support] Firebird replication help needed

[drci...@yahoo.com [firebird-support]]

I am writing a two way server - client Firebird replication with a C# app and I 
am facing the following problem. I have a log table with the table name pk of 
the record and type of the query executet on it. For example if in the sales 
table there has been inserted a line with primery key of 4 there will be the 
table name, pk of the record and insert. My idea is if the statement is insert 
to take every value from the client's DB and insert it into the server's. How 
do I do this and how to do it so that it can work fast. Should I devide every 
record by tables and select all unreplicated records from that table or there 
is some other way. Please help me.
 
 By the way I used the method where every query that has been executed is 
written in a fild in the log table, but that makes it impossible to add columns 
to the table, cos the query will be wrong. Please help me. I've been strugling 
to write a good replication for years.
 

RE: [firebird-support] Firebird replication help needed

['Alan McDonald' a...@meta.com.au [firebird-support]]

Hello Alan,

Thank you for your answer. I will download the FB Replicator try it and see if 
it will work for me. 

That part with the triggers i've already done. I have after I, U, D triggers 
for every table and they insert into my replication log table. The problem is 
that I used to put the sql query that was executedon the client DB on this 
table in the log and then just execute it on the server DB. But if the table 
metadata is changed, then that doesn't work anymore, cos its says can't execute 
to server because the fields in the insert do not equal to those in the table. 
So I decided insted to log only the table name, the PK of record, timestamp and 
the type of query - I U D. And if it's I, to execute insert or update on 
server. But probably when I execute them I'll have to separate them by table 
and get every field name and then value of that table and then onstruct the 
query as a string so it will be probably slow. 

My other problem is should I do it in the C# replication app of use a script 
instead?

Thank you again.If I have further questions I'll write.





There’s IBReplicator (Commercial) at IBPhoenix. FBReplicator is open source.

 

AW: [firebird-support] Support for OS X 10.11 El Capitan

['Paul Beach' pabe...@waitrose.com [firebird-support]]

Having identified and fixed a problem in file_params.h and Info.plist, we
now have a working prototype installer via pkgbuild, albeit very alpha, but 
its installing and working OK for me on El Capitan.

Hopefully I should have something constructive early next week.

Regards
Paul 

AW: [firebird-support] Support for OS X 10.11 El Capitan

[Bernd Frei - ASA bernd.f...@asaon.com [firebird-support]]

Ok, thank you!


Bernd

Von: firebird-support@yahoogroups.com [mailto:firebird-support@yahoogroups.com]
Gesendet: Donnerstag, 8. Oktober 2015 17:14
An: firebird-support@yahoogroups.com
Betreff: AW: [firebird-support] Support for OS X 10.11 El Capitan



Having identified and fixed a problem in file_params.h and Info.plist, we
now have a working prototype installer via pkgbuild, albeit very alpha, but
its installing and working OK for me on El Capitan.

Hopefully I should have something constructive early next week.

Regards
Paul

Re: [firebird-support] Using Zeos 7 and FB 2.5, how do you see in main thread changes made in background thread

[Helen Borrie hele...@iinet.net.au [firebird-support]]

At 02:05 p.m. 9/10/2015, Mark Patterson mark.aurel...@gmail.com 
[firebird-support] wrote:


>Hi
>
>I am using Zeos 7.1.4 stable, and FB 2.5 embedded. I am processing a directory 
>full of SQL scripts to generate a database. All the tables are getting created 
>in a background thread, but after than if I then try to list the tables in the 
>main thread, they don't immediately show up. I have to exit the app and rerun 
>it.
>
>Is there some standard way to have the results of a background thread visible 
>in the main thread?

Commit the transaction in the main thread and start a new one.  Then you will 
see the **committed** work from other transactions.  You will never see 
uncommitted work from other transactions, under any circumstances.

Please would you refrain from unsubscribing and resubscribing to this list 
every time you post a message.  All new subs are moderated - that is to protect 
all of us, including you, from spam.  If you don't want to get the list 
messages, set your account to No Email.

Helen

Re: [firebird-support] Using Zeos 7 and FB 2.5, how do you see in main thread changes made in background thread

[Mark Patterson mark.aurel...@gmail.com [firebird-support]]

On 9 October 2015 at 11:28, Helen Borrie hele...@iinet.net.au
[firebird-support]  wrote:

>
>
> At 02:05 p.m. 9/10/2015, Mark Patterson mark.aurel...@gmail.com
> [firebird-support] wrote:
>
> >Hi
> >
> >I am using Zeos 7.1.4 stable, and FB 2.5 embedded. I am processing a
> directory full of SQL scripts to generate a database. All the tables are
> getting created in a background thread, but after than if I then try to
> list the tables in the main thread, they don't immediately show up. I have
> to exit the app and rerun it.
> >
> >Is there some standard way to have the results of a background thread
> visible in the main thread?
>
> Commit the transaction in the main thread and start a new one. Then you
> will see the **committed** work from other transactions. You will never see
> uncommitted work from other transactions, under any circumstances.
>

Thanks for the suggestion. I wrote this procedure, after a bit of avoiding
exceptions with autocommit mode and the methods:

procedure TdFBZeos.Refresh;
begin
  ZConnection1.AutoCommit := false;
  ZConnection1.Commit;
  ZConnection1.AutoCommit := true;
  ZConnection1.StartTransaction;
end{ Refresh};

I call it as soon as the background thread has fininshed, but I still
cannot see the tables I add until I rerun the program.

Is this what you had in mind?

Regards
Mark

Re: [firebird-support] Using Zeos 7 and FB 2.5, how do you see in main thread changes made in background thread

[Mark Patterson mark.aurel...@gmail.com [firebird-support]]

On 9 October 2015 at 11:52, Mark Patterson  wrote:

> Thanks for the suggestion. I wrote this procedure, after a bit of avoiding
> exceptions with autocommit mode and the methods:
>
> procedure TdFBZeos.Refresh;
> begin
>   ZConnection1.AutoCommit := false;
>   ZConnection1.Commit;
>   ZConnection1.AutoCommit := true;
>   ZConnection1.StartTransaction;
> end{ Refresh};
>
> I call it as soon as the background thread has fininshed, but I still
> cannot see the tables I add until I rerun the program.
>

I do see the new tables I create if I do this in that Refresh method:

  ZConnection1.Disconnect;
  ZConnection1.Connect;

That is a bit drastic, but it will do for this case.

---
Mark

[firebird-support] Using Zeos 7 and FB 2.5, how do you see in main thread changes made in background thread

[Mark Patterson mark.aurel...@gmail.com [firebird-support]]

Hi


I am using Zeos 7.1.4 stable, and FB 2.5 embedded. I am processing a
directory full of SQL scripts to generate a database. All the tables are
getting created in a background thread, but after than if I then try to
list the tables in the main thread, they don't immediately show up. I have
to exit the app and rerun it.


Is there some standard way to have the results of a background thread
visible in the main thread?


TIA
Mark