Re: [sqlite] 3.7.7.1 database corruption

2011-11-14 Thread Filip Navara
Hi Alexandr,

the mailing list doesn't allow attachments. Please send us the output
of pragma integrity_check; on the corrupted database file. Also, is
the database used in WAL mode?

Best regards,
Filip Navara

2011/11/14 Alexandr Němec a.ne...@atlas.cz:
 Dear all,

 we are using SQLite in our projects. We are using the 3.7.7.1 version
 compiled on Windows platform from the amalgamation file. One of our servers
 crashed badly (power outage) and after the server restarted, we saw that the
 database was corrupt (all queries we tried return SQLITE_CORRUPT). We had a
 look at http://www.sqlite.org/howtocorrupt.html and verified, that the
 hardware (disk, memory) of the server should be ok - we used some hardware
 tools to check. In the past we already had some crash situations and SQLite
 could recover from the crash nicely, so we were surprised that after this
 crash the database was corrupt. In fact, we cant create a repro, because we
 do not know, what the server was exactly doing, so it might be difficult to
 find the reason, but we saved the corrupted database in the hope, that
 somebody from the develpers (maybe) could put some light on this problem.
 Because the database is rather small and is used to save some configuration
 data only, I have att
 ached it to this message.

 Thanks for any suggestions.

 Alex



 ___
 sqlite-users mailing list
 sqlite-users@sqlite.org
 http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


Re: [sqlite] 3.7.7.1 database corruption

2011-11-14 Thread Alexandr Němec

Dear Filip,
 
thanks for your reply. I thought that small attachments will go trough. If I run
 
sqlite3 data.db
pragma integrity_check;
 
I receive this message only:
 
Error: database disk image is malformed.
 
Anyway, you can download the database here: http://www.u-consult.cz/db/server.db
 
No, WAL mode is not used.
 
Thanks very much for your help.
 
Alex

__
Od: Filip Navara 
Komu: General Discussion of SQLite Database 
Datum: 14.11.2011 11:17

Předmět: Re: [sqlite] 3.7.7.1 database corruption


Hi Alexandr,

the mailing list doesn't allow attachments. Please send us the output
of pragma integrity_check; on the corrupted database file. Also, is
the database used in WAL mode?

Best regards,
Filip Navara

2011/11/14 Alexandr Němec :

Dear all,

we are using SQLite in our projects. We are using the 3.7.7.1 version
compiled on Windows platform from the amalgamation file. One of our servers
crashed badly (power outage) and after the server restarted, we saw that the
database was corrupt (all queries we tried return SQLITE_CORRUPT). We had a
look at http://www.sqlite.org/howtocorrupt.html and verified, that the
hardware (disk, memory) of the server should be ok - we used some hardware
tools to check. In the past we already had some crash situations and SQLite
could recover from the crash nicely, so we were surprised that after this
crash the database was corrupt. In fact, we cant create a repro, because we
do not know, what the server was exactly doing, so it might be difficult to
find the reason, but we saved the corrupted database in the hope, that
somebody from the develpers (maybe) could put some light on this problem.
Because the database is rather small and is used to save some configuration
data only, I have att
ached it to this message.

Thanks for any suggestions.

Alex



___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users



___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users

___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


Re: [sqlite] 3.7.7.1 database corruption

2011-11-14 Thread Richard Hipp
2011/11/14 Alexandr Němec a.ne...@atlas.cz

 Dear Filip,

 thanks for your reply. I thought that small attachments will go trough. If
 I run

 sqlite3 data.db
 pragma integrity_check;

 I receive this message only:

 Error: database disk image is malformed.

 Anyway, you can download the database here: http://www.u-consult.cz/db/**
 server.db http://www.u-consult.cz/db/server.db


This database file has been truncated.  The last 6 pages (6144 bytes) are
missing.




 No, WAL mode is not used.

 Thanks very much for your help.

 Alex

 __**__**__

 Od: Filip Navara Komu: General Discussion of SQLite Database Datum:
 14.11.2011 11:17
 Předmět: Re: [sqlite] 3.7.7.1 database corruption

  Hi Alexandr,

 the mailing list doesn't allow attachments. Please send us the output
 of pragma integrity_check; on the corrupted database file. Also, is
 the database used in WAL mode?

 Best regards,
 Filip Navara

 2011/11/14 Alexandr Němec :

  Dear all,

 we are using SQLite in our projects. We are using the 3.7.7.1 version
 compiled on Windows platform from the amalgamation file. One of our
 servers
 crashed badly (power outage) and after the server restarted, we saw that
 the
 database was corrupt (all queries we tried return SQLITE_CORRUPT). We had
 a
 look at 
 http://www.sqlite.org/**howtocorrupt.htmlhttp://www.sqlite.org/howtocorrupt.htmland
  verified, that the
 hardware (disk, memory) of the server should be ok - we used some hardware
 tools to check. In the past we already had some crash situations and
 SQLite
 could recover from the crash nicely, so we were surprised that after this
 crash the database was corrupt. In fact, we cant create a repro, because
 we
 do not know, what the server was exactly doing, so it might be difficult
 to
 find the reason, but we saved the corrupted database in the hope, that
 somebody from the develpers (maybe) could put some light on this problem.
 Because the database is rather small and is used to save some
 configuration
 data only, I have att
 ached it to this message.

 Thanks for any suggestions.

 Alex



 __**_
 sqlite-users mailing list
 sqlite-users@sqlite.org
 http://sqlite.org:8080/cgi-**bin/mailman/listinfo/sqlite-**usershttp://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


  __**_
 sqlite-users mailing list
 sqlite-users@sqlite.org
 http://sqlite.org:8080/cgi-**bin/mailman/listinfo/sqlite-**usershttp://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


 ___
 sqlite-users mailing list
 sqlite-users@sqlite.org
 http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users




-- 
D. Richard Hipp
d...@sqlite.org
___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


Re: [sqlite] 3.7.7.1 database corruption

2011-11-14 Thread Alexandr Němec

Dear Richard,
 
huh, thanks very much for the information, I see. But anybody having an idea 
how this can happen? There was a power outage and after restart 6 pages of the 
database file are missing... We checked the disk carefuly, no bad sector etc. 
What can be the reason for this?
 
Thanks once more.
 
Alex
 


__
Od: Richard Hipp 
Komu: General Discussion of SQLite Database 
Datum: 14.11.2011 14:03

Předmět: Re: [sqlite] 3.7.7.1 database corruption

2011/11/14 Alexandr Němec 


Dear Filip,

thanks for your reply. I thought that small attachments will go trough. If
I run

sqlite3 data.db
pragma integrity_check;

I receive this message only:

Error: database disk image is malformed.

Anyway, you can download the database here: http://www.u-consult.cz/db/**
server.db 



This database file has been truncated.  The last 6 pages (6144 bytes) are
missing.





No, WAL mode is not used.

Thanks very much for your help.

Alex

__**__**__


Od: Filip Navara Komu: General Discussion of SQLite Database Datum:
14.11.2011 11:17
Předmět: Re: [sqlite] 3.7.7.1 database corruption

 Hi Alexandr,


the mailing list doesn't allow attachments. Please send us the output
of pragma integrity_check; on the corrupted database file. Also, is
the database used in WAL mode?

Best regards,
Filip Navara

2011/11/14 Alexandr Němec :

 Dear all,


we are using SQLite in our projects. We are using the 3.7.7.1 version
compiled on Windows platform from the amalgamation file. One of our
servers
crashed badly (power outage) and after the server restarted, we saw that
the
database was corrupt (all queries we tried return SQLITE_CORRUPT). We had
a
look at http://www.sqlite.org/**howtocorrupt.htmland verified, that the
hardware (disk, memory) of the server should be ok - we used some hardware
tools to check. In the past we already had some crash situations and
SQLite
could recover from the crash nicely, so we were surprised that after this
crash the database was corrupt. In fact, we cant create a repro, because
we
do not know, what the server was exactly doing, so it might be difficult
to
find the reason, but we saved the corrupted database in the hope, that
somebody from the develpers (maybe) could put some light on this problem.
Because the database is rather small and is used to save some
configuration
data only, I have att
ached it to this message.

Thanks for any suggestions.

Alex



__**_
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-**bin/mailman/listinfo/sqlite-**users


 __**_

sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-**bin/mailman/listinfo/sqlite-**users


___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users





--
D. Richard Hipp
d...@sqlite.org
___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users

___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


Re: [sqlite] 3.7.7.1 database corruption

2011-11-14 Thread Kees Nuyt
On Mon, 14 Nov 2011 14:16:46 +0100, Alexandr N?mec
a.ne...@atlas.cz wrote:

 Dear Richard,
 
 huh, thanks very much for the information, I see.
 But anybody having an idea how this can happen?
 There was a power outage and after restart 6 pages
 of the database file are missing...
 We checked the disk carefuly, no bad sector etc.
 What can be the reason for this?

The disk or the filesystem may have confirmed a write or sync
while in fact the data wasn't on disk yet.
Maybe the filesystem metadata is updated in a separate
transaction, or your filesystem doesn't journal metadata.
fsck only repairs inconsistencies it can find, depending on teh
filesystem used there are quite a few corruptions it cannot detect
nor repair.

Did you use any dangerous PRAGMA to speed things up, like running
without database journals? 

Does some startup script remove journal files before the database
is connected to by an application?
-- 
  (  Kees Nuyt
  )
c[_]
___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


Re: [sqlite] 3.7.7.1 database corruption

2011-11-14 Thread Alexandr Němec

Thanks again for your answer,
 
no, no dangerous pragmas, no journal file removing. Our file system is NTFS so 
I hope that no stupid things are done by the file system :). In fact, we 
changed just two things only in the amalgamation file - we increased 
SQLITE_MAX_PAGE_COUNT to 2 000 000 000 and are using SQLITE_THREADSAFE 2 - we 
are using multiple database files from inside one application, but there are 
separate database connections for each file and every connection is accessed by 
a single thread only (but two queries for two different files using two 
different connections can run simultanously in our app). So I hope we are 
correct here. We tested a lot of crash situations in the past and never 
experienced a problem. Now we were very surprised about the missing pages.
 
Thanks.
 
Alex

__
Od: Kees Nuyt 
Komu: 
Datum: 14.11.2011 15:52

Předmět: Re: [sqlite] 3.7.7.1 database corruption


On Mon, 14 Nov 2011 14:16:46 +0100, Alexandr N?mec
wrote:


Dear Richard,
 
huh, thanks very much for the information, I see.
But anybody having an idea how this can happen?
There was a power outage and after restart 6 pages
of the database file are missing...
We checked the disk carefuly, no bad sector etc.
What can be the reason for this?


The disk or the filesystem may have confirmed a write or sync
while in fact the data wasn't on disk yet.
Maybe the filesystem metadata is updated in a separate
transaction, or your filesystem doesn't journal metadata.
fsck only repairs inconsistencies it can find, depending on teh
filesystem used there are quite a few corruptions it cannot detect
nor repair.

Did you use any dangerous PRAGMA to speed things up, like running
without database journals? 


Does some startup script remove journal files before the database
is connected to by an application?
--
 (  Kees Nuyt
 )
c[_]
___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users

___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


Re: [sqlite] 3.7.7.1 database corruption

2011-11-14 Thread Kees Nuyt
On Mon, 14 Nov 2011 18:24:46 +0100, Alexandr N?mec
a.ne...@atlas.cz wrote:

Thanks again for your answer,
 
 no, no dangerous pragmas, no journal file removing.
 Our file system is NTFS so I hope that no stupid
 things are done by the file system :). In fact,
 we changed just two things only in the amalgamation file
 - we increased SQLITE_MAX_PAGE_COUNT to 2 000 000 000
  and are using SQLITE_THREADSAFE 2
 - we are using multiple database files from inside one
 application, but there are separate database connections
 for each file and every connection is accessed by a single
 thread only (but two queries for two different files using
 two different connections can run simultanously in our app).
 So I hope we are correct here.

That all sounds healthy to me.

 We tested a lot of crash situations in the past and never
 experienced a problem.
 Now we were very surprised about the missing pages.

I can image. Soryy, I have no more ideas.

Thanks.
 
Alex

__
 Od: Kees Nuyt 
 Komu: 
 Datum: 14.11.2011 15:52
 P?edm?t: Re: [sqlite] 3.7.7.1 database corruption

On Mon, 14 Nov 2011 14:16:46 +0100, Alexandr N?mec
 wrote:

 Dear Richard,
 
 huh, thanks very much for the information, I see.
 But anybody having an idea how this can happen?
 There was a power outage and after restart 6 pages
 of the database file are missing...
 We checked the disk carefuly, no bad sector etc.
 What can be the reason for this?

The disk or the filesystem may have confirmed a write or sync
while in fact the data wasn't on disk yet.
Maybe the filesystem metadata is updated in a separate
transaction, or your filesystem doesn't journal metadata.
fsck only repairs inconsistencies it can find, depending on teh
filesystem used there are quite a few corruptions it cannot detect
nor repair.

Did you use any dangerous PRAGMA to speed things up, like running
without database journals? 

Does some startup script remove journal files before the database
is connected to by an application?
-- 
  (  Kees Nuyt
  )
c[_]
___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


Re: [sqlite] 3.7.7.1 database corruption

2011-11-14 Thread Kees Nuyt
On Mon, 14 Nov 2011 21:03:56 +0100, Kees Nuyt k.n...@zonnet.nl
wrote:

I can image. Soryy, I have no more ideas.

Oops. Make that:
I can imagine. Sorry, I have no more ideas.
-- 
  (  Kees Nuyt
  )
c[_]
___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


[sqlite] 3.7.7.1 database corruption

2011-11-13 Thread Alexandr Němec

Dear all,
 
we are using SQLite in our projects. We are using the 3.7.7.1 version compiled 
on Windows platform from the amalgamation file. One of our servers crashed 
badly (power outage) and after the server restarted, we saw that the database 
was corrupt (all queries we tried return SQLITE_CORRUPT). We had a look at 
http://www.sqlite.org/howtocorrupt.html and verified, that the hardware (disk, 
memory) of the server should be ok - we used some hardware tools to check. In 
the past we already had some crash situations and SQLite could recover from the 
crash nicely, so we were surprised that after this crash the database was 
corrupt. In fact, we cant create a repro, because we do not know, what the 
server was exactly doing, so it might be difficult to find the reason, but we 
saved the corrupted database in the hope, that somebody from the develpers 
(maybe) could put some light on this problem. Because the database is rather 
small and is used to save some configuration data only, I have att
ached it to this message.
 
Thanks for any suggestions.
 
Alex


___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users