System.Data.SQLite

Hi Robert,

I have a user who's getting "The database disk image is malformed" errors occasionally. I did a search and didn't find anything about how this can happen. Might you know?

He sent me the database file and the Sqlite3 CLI does give the same error when I try to inspect the database. I can forward it if it would help. I'm using v1.0.38.

Thanks!

Sorry for not getting back to you sooner.  If the program crashed while a database operation was going, and the journal file was deleted before the app was restarted, it could cause an error like that. 

Robert

 

hi,

i have same problem,and i understand why this problem get.but i dont know how  to solve?please help me!

Are you using encryption?  Can you tell me any steps or give me any information about your environment?

I have at least two customers seeing this.  

I am not using encryption.

One particular table has all queries fail with errors such as the database disk image is malformed. I see this in Sqliteadmin when the customer sent me the .db3 file.

There was no .journal file that he could find. It was stored in main memory, not an SD card.

I can view all the schema in the file, but one particular table always errs out if I try to check the data.

Customer wants to know if the data is recoverable. I don't know how to recover from a corrupted sqlite file.

I have seen this error occuring under the following conditions:

I am using pure SQLite on Windows Mobile 5 devices, databases are synced automatically via ActiveSync 4.5. While sync is in progress the user removes the PDA from the cradle. This leaves the version on the PC intact but not the PDA-database (it is truncated). On next sync the defective db is copied to the PC, oh my.

Another incident which was observed once:
The PDA (iPAQ hx2490) was cradled with a program running having the database open. Overnight the PDA somehow managed to discharge the battery so completely, that the battery refused to charge again. Several attempts and soft reboots were necessary to revive the battery. After that the db was truncated to just 1/3 of the size it should have been. It was unclear whether the PDA itself damaged it or ActiveSync later.

These kinds of problems seem to have nothing to do with SQLite or the wrapper.

Although I dont' know it for sure in this case, synchronization issues seem likely. They've happened in the past. I think you can setup folders to Autosync semi continuously, and I think that is a very bad idea.

Are there ways of attempting to recover some or all of the data? I've been looking at in SQLiteAdmin, and so far I don't know.

 

 

I think it's not a good idea to rely on ActiveSync alone. The solution depends on your needs.

In our project we handle it as following:
On the PDA we have 2 databases. One with master data (articles, suppliers etc.). It originates from the PC, does not change on the PDA and is easily copied in case of failure. The other db contains data which is collected on the PDA several times a day. After a collection the PDA is cradled and the data are transferred to the PC for further processing. On the PC we keep a copy of those data, so in case of a db-destruction on the PDA it can be reconstructucted and transferred to the PDA again.

BTW: We made the experience, that Windows Mobile devices do not have the stability and reliabilty your are familiar with on Windows XP.

I met this database disk image is malformed on an encrypted database. Only a particular table is affected. Are today some workarounds?

I figured it out it was caused by a DROP TRIGGER statement executed in a Transaction. Robert please look into this problem and fix it for the next release.

Not a wrapper issue ... this is an engine issue.

Are you sure? Have you met this before? Can't you make the wrapper to not execute a trigger drop if runs inside in a transaction? Give proper error.

No I sure can't.  Everyone thinks the wrapper is doing a lot of work, but it's really not.  The wrapper is a a very lightweight interface to the core engine.  I don't read, parse, or attempt to make sense of the commands you are sending to the core engine as that would be a significant performance hit.  I just pass your queries to the core engine verbatim.

I make no attempt to figure out what you're doing or why you're doing it when you execute commands.

Just wanted to share some info. I am having the exact same error with a different wrapper. When I try to drop the trigger, I get an error that the trigger does not exist, but when I try to create the trigger after the drop, I get an error message that the trigger already exists. At that point my database is malformed and completely corrupt. I suspect that I am using the wrong DLL function within the wrapper. My wrapper is routing all SQL commands (other than open, close, free memory type stuff) through sqlite3_get_table. I have sent a note to the SQLITE users list for assistance. I hope to find an alternative function that will execute the drop trigger and/or create trigger without crashing the data. I don't know if any of this does or does not apply to your situation, but the symptoms seemed way too similar to ignore. John

Actually I was able to fix this problem. It turns out this happened because it was issued in a transaction with some other commands. And SQLite doesn't like manipulating triggers in a transaction. Once I moved out the trigger of the transaction it worked ok. Maybe you should look into, and see if this is the situation.