Hi, All.

I use Firebird 0.9-4p1 on Linux RH 6.2. Server is powered by UPS, no
user have file access to .gdb, all connection strings are identical,
.gdb size is about 2Gb (few smaller), secondary file created but not
used yet. Sweep is turned off, every night system reboots with network
turned off, make file copy of .gdb, perform backup and sweep and
reboots with network turned on.
This night backup was not successful, and logged error is
internal gds consistency check (Decompression overrun buffer (179))
Making backup on copy of .gdb with -verbose showed that error
occures when backing up data from one of the tables.
Gfix -v -f finished with message
Summary of validation errors
Number of record level 3
After gfix -mend I could make .gbk but restore is not successful -
it stopped when restoring the same table and encountered Null value in
column specified as Not Null.
After all, users for a while does not yet noticed any troubles with
routine dayli work.

I have three questions.
1. Why it happened and what I must do to prevent it in future.
2. How can I repair .gdb to state when I can do backup/restore.
3. How can I determine what data is lost after 2.

Thank you for advices and comments.

Alexander V. Nevsky.