""Richard"" <rad@...> escreveu na mensagem
news:bf8do8+bchh@......

> Most of our Customers use FB 1.0. And until now we have had about 5
> demaged databases that we have to backup und restore to repair the
> erros.
> Some problems are that some entries are not deleted but without any
> error-message prompted for the user (we are checking if our Zeos-
> Components dont work properly). But there are also no entries in the
> interbase.log that show me a problem within the database until i

make

> a validation of the DB.

Yesterday (17-JUL-2003) . The users reported that they didn't find the
registry of patients that the knew that was in the hospital.
I've shutted the database down. I've checked the log, and no message
was
corruption error. Just the usual "INET/inet_error: read errno = 104".
Whe I issued a gfix -v -full, I started to see index page erros in
the log.

Fortunately after a backup and restore my data was intact. (3 hours of
downtime)

FireBird 1.0.2 Classic 64IO 1
2.GB database
One active Shadow in another hd (local partition)
Forced writes : On
Linux kernel 2.4.19
File system reiserfs (partition with sync)
UPS

During the downtime I've upgrade to 1.0.3 , because it says that "a
couple
of buffer overuns" was fixed.

Have anyone encountered a situation like this ?

Thanks in advance,

Geraldo Lopes de Souza