Hi folk When I run a normal routine for database checking, the database passes. When I run the exhaustive testing, (with all option flagged) the database doesnt pass. Is that normal behavour. /Frank
I forgot to tell that its an Attain 3.6 database split into tree parts.
Which test is giving the error?
I will have to get on that. I think it was the secondary key test - but I am not certain. I will get back on that /FL
If its a secondary key test, then the data is OK, but you have a problem that must be rectified.
Ok … And how is that done … Any tools? /FL
The only way to fix, is a backup and restore (assuiming its a secondary key), but that’s the easy part. This is normally the resiult of a hardware failure, but it is very very hard to find out what failed. I have seen this error enough times to know that locating the source is a big job. Your first thing is to be aware, that if the area of disk where the secondary key is stored becomes available, and is then used by a primary key, the whole database could crash, and then it can only be recovered using C/DART. So I would be making very regular backups at the moment till you get it sorted. And ak sure to make Navision backups, so that you have primary key integrity. The two most common issues are Network problems. And Harddrive issues. (Though many people get stuck on Harddrives, and never look at anything else). I have seeen mor entwork issues cause this than drive failures. The most common cause is a lower performance client that is doing some complex task involving writing to the database, that has a faulty network card, and is running at the same time as a number of other large tasks such as backuup or inventory adjustments. You should get a network analyer and monitor bad packets. Running Inventory adjustments, backups large reports etc, to generate a lot of traffic. (Of course make a backup first). Also do a thourough scan of all your drives. Preferable seperating the RAID1’s into seperate drives for the testing. Just be prepared that the system could fail, so keep those backups moving. PS, I don’t even want ot think this, but if for some bizare reason you have your database on a RAID 5, the reason would then be obvious, but then no one would use RAID 5 with Navision, so we know that can’t be the issue.
Nope - No Raid 5. But unfortunately the databaseparts is put into 3 parts. The space for making these 3 parts, is done by breaking the mirror, in a way that its generating errors. The disks are now being used as oridinary non mirrored disk. They are still marked as mirrored in Disk Managment, but the mirror is broken and now have different content in them. So my first and most important task is to have the content backed up, and the disk array fixed, and then restored. Then I could make a separate Navision server solely build up for testing. There I would try to use different tools on a backup database. When I am successfull in the repair, I could try it on a contempory copy of the database, and if successfull use that database as my current database. Does it sound like wise path or what? PS. Thank you for your prompt answers by the way !
Looks like good news then, becasue pretty certainly that is the problem. How big is the database? Make a backup, get rid o the old drives, install a new set of RAID 1 arrays, restore the database. Hard disks are just too cheap to waste time on such experimenting. I am sure you will find another use for the drives. I am assuming that you have dediicated Navision server. Another option would be to buid a new server completely, and then use the old one as your test machine.
The database is split into 3 parts of each 1.9 GB Which tools could be applied to cure the database?
The way I understand David, do the following: Make a backup of your Database (Tools\Backup), create a new database on a different Server, Restore the Backup into that Database. Regards Daniel
Exactly. Basically 2 G per Raid 1 arry is fine, in fact 4G is no problem these days. It would be best to install a whole new server, but at the minimum, install three new RAID 1 arrays. If the server does crash its going to cost you 2-3 days of consulting, which is much more expensive that a new machine. Not to even mention the down time.
Will such a backup and restore effectively cure the problem with de missing references? /Frank