[GRASS-user] sqlite-db broken?

[Markus Metz]

Hi,

On Mon, Jul 16, 2012 at 4:02 PM, Johannes Radinger <JRadinger at gmx.at> wrote:
> Hi,
>
> after some search in the internet, it seems
> the sqlite doesn't work very well with samba-network
> drives (mounted via cifs). There is somehow
> an issue with the locking of the db.

I think I found the reason [0]:

"When SQLite creates a journal file on Unix, it opens the directory
that contains that file and calls fsync() on the directory, in an
effort to push the directory information to disk."

If another process is modifying files or just keeping files open in
this directory while sqlite calls fsync(), opening the sqlite database
will fail.

Markus M

[0] http://www.sqlite.org/lockingv3.html

>
> Does anyone have experience with using GRASS/SQLITE
> on a cifs-mounted GRASS location?
>
> What I found is following mount-option:
>
> nobrl
> Do not send byte range lock requests to the server. This is necessary for certain applications that break with cifs style mandatory byte range locks (and most cifs servers do not yet support requesting advisory byte range locks).
>
> But I am not really sure if this will work or even destroy something.
> So I'll let that judge some expert of you!
>
> cheers,
>
> Johannes
>
> -------- Original-Nachricht --------
>> Datum: Mon, 16 Jul 2012 14:14:48 +0200
>> Von: "Johannes Radinger" <JRadinger at gmx.at>
>> An: "Johannes Radinger" <JRadinger at gmx.at>, grass-user at lists.osgeo.org
>> Betreff: Re: [GRASS-user] sqlite-db broken?
>
>> Hi again,
>>
>> I did another test now and found out following:
>>
>> It seems that the problem with the sqlite database
>> might be related to the directory of my grass location.
>> When I try my to save the location on my local drive
>> everything works, there is just a problem when
>> my location points to a directory which is actually a mounted
>> samba network drive (mount cifs).
>>
>> Here an output of the directory which is actually a network drive:
>> radinger at grassgis:~$ ls -la
>> /home/radinger/U_Radinger/05_GRASS/GRASS_locations/
>> total 0
>> drwxrwxrwx 5 radinger radinger 0 Jul 16 12:01 .
>> drwxrwxrwx 6 radinger radinger 0 Jul 13 14:09 ..
>> drwxrwxrwx 4 radinger radinger 0 Jul 16 11:43 xxx
>> drwxrwxrwx 4 radinger radinger 0 Jan 26  2009 gisdemo_ncspm
>> drwxrwxrwx 4 radinger radinger 0 Jul 12 14:16 spearfish60
>> drwxrwxrwx 3 radinger radinger 0 Jul 16 10:44 xxx
>>
>> and:
>> radinger at grassgis:~$ ls -la
>> /home/radinger/U_Radinger/05_GRASS/GRASS_locations/gisdemo_ncspm/sqlite/
>> total 6
>> drwxrwxrwx 3 radinger radinger   0 Jul 16 13:15 .
>> drwxrwxrwx 4 radinger radinger   0 Jan 26  2009 ..
>> -rwxrwxrwx 1 radinger radinger  24 Jul 16 12:08 .bash_history
>> -rwxrwxrwx 1 radinger radinger 556 Jul 16 12:03 .bashrc
>> -rwxrwxrwx 0 radinger radinger 512 Jul 16 12:04 cifs2e55
>> -rwxrwxrwx 1 radinger radinger   0 Jul 16 12:03 sqlite.db
>> drwxrwxrwx 2 radinger radinger   0 Jul 16 12:02 .tmp
>> -rwxrwxrwx 1 radinger radinger  74 Nov  3  2008 VAR
>> drwxrwxrwx 1 radinger radinger   0 Nov  3  2008 vector
>> -rwxrwxrwx 1 radinger radinger 361 Nov  3  2008 WIND
>>
>>
>> Does anyone know hat causes the problem exactly and what should I change?
>>
>> The db-connection is set to $GISDBASE/$LOCATION_NAME/$MAPSET/sqlite.db
>> (checked in the VAR file as well)
>> Maybe the GRASS can't tranlate from the relative path to a hardcoded one
>> when the $GISDBASE is already located on a mounted network drive?
>>
>> /Johannes
>>
>> -------- Original-Nachricht --------
>> > Datum: Mon, 16 Jul 2012 12:04:40 +0200
>> > Von: "Johannes Radinger" <JRadinger at gmx.at>
>> > An: grass-user at lists.osgeo.org
>> > Betreff: [GRASS-user] sqlite-db broken?
>>
>> > Hi,
>> >
>> > I used to use db.test (with test=test1) to test my sqlite database
>> > connection. In the most recent verson of GRASS 6.4.3 (rev 52380) it
>> seems to be
>> > broken. I tried it on some of my own databases and a freshly downloaded
>> > sample dataset (small NC dataset, mapset=sqlite). I started it from the
>> gui but
>> > it just get stucked in the process (no output after several
>> > minutes)...stopping it leaves the database in an incosistent state.
>> >
>> > Thats why I tried just to simply copy a vector file (e.g streams) from
>> > PERMANENT to sqlite. Then I get the error message: Unable to open header
>> file
>> > of vector <streams at sqlite> Unable to read head file....
>> >
>> > Somehow working with sqlite doesn't work properly on my machine (Ubuntu
>> > 12.04) although I the command sqlite3 in the terminal works....
>> >
>> > I also tried it with 6.5 SVN (rev. 52380) and a fresh NC dataset again:
>> > db.test just gives: Busy SQLITE db, already waiting for 10 seconds...
>> >
>> > What else do I need to test so that someone of you might know the
>> problem
>> > of my basic database functions (creating tables eg. when tranforming
>> from
>> > raster to vector etc.)? Can anyone reproduce this or is it just a
>> behavior
>> > due to my own compilation (configured with sqlite3 support)?
>> >
>> > /Johannes
>> > _______________________________________________
>> > grass-user mailing list
>> > grass-user at lists.osgeo.org
>> > http://lists.osgeo.org/mailman/listinfo/grass-user
>>
> _______________________________________________
> grass-user mailing list
> grass-user at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/grass-user