Subject | Advise on setting flags for backup and restore |
---|---|
Author | GOVINDKRISHNA |
Post date | 2003-01-22T09:59:35Z |
Firebird 1.0 using IBO and TIBBackupService,TIBRestoreService and TIBValidateService
I am shipping an application which goes out into the wild.
Though i strongly recommend that the customers install the server on NT or 2K machines there are some who do install it on a Windows 98 machine.
I am not modifying the setting for Forced Write.
Database gets corrupted occasionally. Almost all the cases the customer does confirm that the machine did crash just before the corruption
Always on Win 98. Almost all cases it is a checksum error on blob files.
If I do a validate and backup and restore ignoring errors the rest of the database comes up.
But after that I am just not able to do a backup without ignore errors flag set.
So my application ships with backup with ignore errors flag set. This is making me very uncomfortable.
I try to make my applications zero administration as far as possible.
So my program has an automatic backup setting. Default once a day
It also has what I call an automatic housekeeping setting. Default once in 30 days.
Here I prevent clients from connecting do a full backup and restore
Right now my flag settings on the backup is
IgnoreChecksums,NonTransportable
and the flag setting for restore is
NoShadow,NoValidityCheck,OneRelationAtATime,CreateNewDB,Replace
Any suggestions/ recommendations on what should be the flags for this kind of usage.
This is a very low cost app and I am willing to put in any kind of effort when it comes to development but I really want to make support at customer location ZERO.
Also I should mention that what I tried initially was to take a backup without setting the IgnoreCheckSums flags and trap an error and start another backup with IgnoreCheckSum Flags set. However if there is an error during backup I must restart the machine before I can actually start another backup. I check no of attachments to the database and it is zero still the backup cannot be initiated again.
Any suggestions appreciated
GK
I am shipping an application which goes out into the wild.
Though i strongly recommend that the customers install the server on NT or 2K machines there are some who do install it on a Windows 98 machine.
I am not modifying the setting for Forced Write.
Database gets corrupted occasionally. Almost all the cases the customer does confirm that the machine did crash just before the corruption
Always on Win 98. Almost all cases it is a checksum error on blob files.
If I do a validate and backup and restore ignoring errors the rest of the database comes up.
But after that I am just not able to do a backup without ignore errors flag set.
So my application ships with backup with ignore errors flag set. This is making me very uncomfortable.
I try to make my applications zero administration as far as possible.
So my program has an automatic backup setting. Default once a day
It also has what I call an automatic housekeeping setting. Default once in 30 days.
Here I prevent clients from connecting do a full backup and restore
Right now my flag settings on the backup is
IgnoreChecksums,NonTransportable
and the flag setting for restore is
NoShadow,NoValidityCheck,OneRelationAtATime,CreateNewDB,Replace
Any suggestions/ recommendations on what should be the flags for this kind of usage.
This is a very low cost app and I am willing to put in any kind of effort when it comes to development but I really want to make support at customer location ZERO.
Also I should mention that what I tried initially was to take a backup without setting the IgnoreCheckSums flags and trap an error and start another backup with IgnoreCheckSum Flags set. However if there is an error during backup I must restart the machine before I can actually start another backup. I check no of attachments to the database and it is zero still the backup cannot be initiated again.
Any suggestions appreciated
GK