application ends suddently without show any error 1

[gcamps]

I have a big Clipper 5.2d Application linked by blinker 4.1 and I have a problem, it is a network app running on a w98 server (over w2000 I have the same problem), when the server program is in a loop checking for updates between 2 dbfs A.dbf related with B.dbf by IDProd field and any client machine make a select statement then set order , and then try to move the a.dbf pointer, the server application ends suddently without show any error, just exit. the same problen appears some times when client machine make a commit, I tryied with different net protocols, and operating systems blinker 5 and the problem continue, if I run to dos sessions at the server doing the same test everything works fine.
thank you for any suggestion, and sorry for my english

Gustavo Camps - Cordoba - Argentina

 

[Bozz]

Gustavo,

I think that the patch up to 5.2e is available freely on the CA site and it's generally thought to be the most stable and bug free, so it might be worth upgrading.

It sounds to me like Internal error 19, a corrupt index. It might be worth looking at the DOS box properties and stop it from closing automatically, then you may see the error. Or, if you run it from a batch file, put a pause after the exe.

Rebuilt the indexes (rather than reindex as this may not get rid or all problems) and it might be OK.

Good Luck

Ian Boys
DTE Systems Ltd

 

[Indio]

Te contesto en español:
Modifica ligeramente tu programa, de manera que no cierre los archivos al pasar de uno a otro. Dejalos abiertos y solo haz SELECT entre uno y otro.
Eso seguramente solucionara tu problema
(Si se soluciona, avisame a e200000@hotmail.com)

 

[AnanthaP]

I don't think that it is a problem with protocols. You are clearly identifying the instance to a client process that "disturbs" the sequence of the server process. Also, sometimes, it fails during commit.

Quote
same problen appears some times when client machine make a commit
UnQuote

So how to solve it?

Please check the "client" application. It is probably taking exclusive lock on the a.dbf table. Thus the server app is abending.

As a first step, please introduce neterr() in the server app. At least, it will allow you to exit with message and give a clue.

end