suhas-mahajan
26th February 2003, 09:49
Dear *.*
Is there any alternate of bdbpre? My ultimate aim is how to take sequential backup without locking the table?
-Suhas
norwim
26th February 2003, 10:14
yes, the lock that bdbpre creates is really annoying.
OTH bdbpre usually is very fast, so if you split up the job and run several bdbpre jobs parallelly (?spelling?), only a short timespan is needed where nobody should work.
But if you want to dump tables without locking them, you can either write a Baan program that uses seq-commands to write to files or
create a script to be run by oracle to create a dump-file (although I don't have any ideas whether this will work without the table to be locked by oracle itself).
hth
Norbert
suhas-mahajan
26th February 2003, 10:42
Dear Norwin,
My company works 7 * 24 hrs. and not accepts any down-time. So short timespan not possible.
I tried for seq.open way, but how to make, a text stream with all the columns with "|" separator, I am not able to do/search it.
My oracle side is not that much stronger, so I couldn't do it.
Any more idea.
-Suhas
patvdv
26th February 2003, 11:10
Suhas,
If your company runs an 'always-on' shop then you should really looking into taking online or hot database backups. Oracle provides specific tools for this (e.g. RMAN).
suhas-mahajan
26th February 2003, 11:23
Dear Respected Sir,
I am surprised/shocked/impressed, guru's guru replying me.
Sir, we are taking cold-backup, hot-backup, as well as my database is on Archive mode (online-backup) with 2 hrs. replication distance.
I am using bdbpre or ttaad4226m000 for some other issue. i.e. updating my MS-Access base site, In VB, I made a program which converts this .S files into .TXT and from .TXT to .MDB.
But because of night shift frequntly, we couldn't activate job. Is there any way?
-Suhas
patvdv
26th February 2003, 11:37
LOL Suhas,
You can skip the divine terms :)
Anyway, you described your problem more clearly now. The only way I can see this happening for you is that you run a logical (oracle) export of your Baan company by supplying a list of the company's tables to the 'exp' command. However, that would leave you with a dump in a propietary format. Not sure if there are any available ways to convert Oracle dumps into any readable formats.
suhas-mahajan
26th February 2003, 12:30
Dear Respected Sir,
I can use exp command of Oracle, but for that I have to change entire architecture of my site, which will big task for me.
That's why I searching alternate of bdbpre.
-Suhas
norwim
26th February 2003, 12:41
it is not that complicated to create sequential files with a baan program.
I even wrote a little program that creates the program script more or less automagically. This program should be on the board here. There are some drawbacks (the pvrc has to be modified in the script, texts are not dumped), but it should give you an idea.
What is basically does is go through the tools tables to find out which fields (of what kind) are in the table and then creates a template that can be used to generate a baan program to dump the whole file into a sequential one.
These programs are very fast too, we dump about 30 tables each night within < 20 minutes. (These files are then imported into our mysql DB of our intranet)
If you give me your email adress, I will mail you this sample, if you can't find it here.
regards
Norbert
suhas-mahajan
26th February 2003, 13:05
Dear Norbert,
My e-mail ID is : suhas-mahajan@indiatimes.com
I have searched codes related to bdbpre found following by Respected NP Rao. But it again internally uses bdbpre, Is this that code, which you are talking about?
--------------------------------------------------------------------------
select ttadv420.cpac, ttadv420.cmod, ttadv420.flno
from ttadv420
where ttadv420.cpac <> "tt"
and ttadv420.expi = ttyeno.no
and ttadv420._compno = 000
group by ttadv420.cpac, ttadv420.cmod, ttadv420.flno
selectdo
ret = change.progress.indicator(0, ttadv420.cpac &
ttadv420.cmod & ttadv420.flno )
tab.name = ttadv420.cpac & ttadv420.cmod & ttadv420.flno
comd = "bdbpre" & bse.release$() & " -t""|"" -N" & tab.name & " -o " & path & " -C" & comp
e = shell(comd, 0)
endselect
--------------------------------------------------------------
Waiting for your reply.
-Suhas
norwim
26th February 2003, 13:54
No, you can find the script under "table loader" here on the board.
feel free to ask me
wimmer@rug.de or
now@niederrhein.de
if you have troubles to get it to work
regards
Norbert
nick_rogers
3rd March 2003, 18:01
what about Exchange to create a SEQ file for the table(s) ?
askajale
3rd March 2003, 20:02
I can suggest another way of doing it.
Write and Oracle SQL script (with all heading off, line size and all other parameters to set) which will generate the text file. Now to get the "|" seperator, you can say select tdsls040.orno, "|", tdsls040.pono, .... from... where ...
It may be difficult to write this sql, but what we had done (at one of our client) is to generate the dynamic sql script from the dict folder of the package combination using unix shell and awk script.
This sql extract will not lock the table and on performance it is really very fast.
But ofcourse there are some huge initial efforts.
-- Avinash