Обсуждение: Is my db dead ?

My server crashed yesterday, the hd was saved though and I was able to
get a copy of the data directory of my pg cluster. On the server now,
when I start pg, I can see only like 1% of my data, there are many dbs/
tables and rows missing. But the data dir seems ok in size though. So
I copied the whole data dir to a new serv, installed the same pg
version but now I can't get it started at all. There are some weird
things in the pg_controldata :

pg_control last modified:             Tue 03 Sep 6869 17:57:36 CET
Time of latest checkpoint:            Thu 11 Jan 1116953 03:46:31 CET
LC_COLLATE:
LC_CTYPE:

So now I can't start the server because of the blank locale :

Starting PostgreSQL 8.1 database server: main* Error: The server must
be started under the locale : which does not exist any more.

Soo... Is there a chance for me to see again my data ? What can I do
to fix this ?
Thanks for any suggestions

dawmette@gmail.com escribió:
> My server crashed yesterday, the hd was saved though and I was able to
> get a copy of the data directory of my pg cluster. On the server now,
> when I start pg, I can see only like 1% of my data, there are many dbs/
> tables and rows missing. But the data dir seems ok in size though. So
> I copied the whole data dir to a new serv, installed the same pg
> version but now I can't get it started at all. There are some weird
> things in the pg_controldata :
>
> pg_control last modified:             Tue 03 Sep 6869 17:57:36 CET
> Time of latest checkpoint:            Thu 11 Jan 1116953 03:46:31 CET
> LC_COLLATE:
> LC_CTYPE:

What's the full pg_controldata output?  Can you send the pg_control file
as a binary attachment?

--
Alvaro Herrera                                http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support

Ok in fact, the times weird thing were because I copied the data dir on a 64 bits arch and it went from a 32 bit arch. On the 32 bit one :
pg_control version number:            812
Catalog version number:               200510211
Database system identifier:           4989617165674917617
Database cluster state:               in production
pg_control last modified:             Mon 04 Feb 2008 20:43:33 CET
Current log file ID:                  0
Next log file segment:                34
Latest checkpoint location:           0/2109832C
Prior checkpoint location:            0/210982E8
Latest checkpoint's REDO location:    0/2109832C
Latest checkpoint's UNDO location:    0/0
Latest checkpoint's TimeLineID:       1
Latest checkpoint's NextXID:          1315262
Latest checkpoint's NextOID:          34543
Latest checkpoint's NextMultiXactId:  1
Latest checkpoint's NextMultiOffset:  0
Time of latest checkpoint:            Mon 04 Feb 2008 20:05:40 CET
Maximum data alignment:               4
Database block size:                  8192
Blocks per segment of large relation: 131072
Bytes per WAL segment:                16777216
Maximum length of identifiers:        64
Maximum columns in an index:          32
Date/time type storage:               64-bit integers
Maximum length of locale name:        128
LC_COLLATE:                           fr_FR.utf8
LC_CTYPE:                             fr_FR.utf8

Everything seems ok but I still can't see 99% of my data. No error msg, nothing special, just a ton of relations does not exist and empty or almost empty tables.

Ok I progressed a bit, now with psql I can see my main db and some tables, but they don't contain many rows.
Noticable messages in the logfile :
2008-02-05 18:01:18 CET LOG:  transaction ID wrap limit is 1073748480, limited by database "wowdbu"
2008-02-05 18:01:18 CET LOG:  autovacuum: processing database "postgres"
2008-02-05 18:01:18 CET LOG:  incomplete startup packet
2008-02-05 18:02:08 CET LOG:  incomplete startup packet

I tried a pg_dumpall and it fails quickly with this error :
pg_dump: SQL command failed
pg_dump: Error message from server: ERROR:  could not access status of transaction 3270404
DETAIL:  could not open file "pg_clog/0003": No such file or directory
pg_dump: The command was: SELECT tableoid, oid, proname, prolang, pronargs, proargtypes, prorettype, proacl, pronamespace, (SELECT rolname FROM pg_catalog.pg_roles WHERE oid = proowner) as rolname FROM pg_proc WHERE NOT proisagg AND pronamespace != (select oid from pg_namespace where nspname = 'pg_catalog')
pg_dumpall: pg_dump failed on database "postgres", exiting

I read about things like zeroing missing pg_clog files... Can it work ? I thought clog files were about transactions, but I don't care about last transactions, even data from 1 month ago would be enough for me.
So what can I do from there ?

Eric Renard escribió:

> pg_dump: Error message from server: ERROR:  could not access status of
> transaction 3270404
> DETAIL:  could not open file "pg_clog/0003": No such file or directory
> pg_dump: The command was: SELECT tableoid, oid, proname, prolang, pronargs,
> proargtypes, prorettype, proacl, pronamespace, (SELECT rolname FROM
> pg_catalog.pg_roles WHERE oid = proowner) as rolname FROM pg_proc WHERE NOT
> proisagg AND pronamespace != (select oid from pg_namespace where nspname =
> 'pg_catalog')
> pg_dumpall: pg_dump failed on database "postgres", exiting
>
> I read about things like zeroing missing pg_clog files... Can it work ? I
> thought clog files were about transactions, but I don't care about last
> transactions, even data from 1 month ago would be enough for me.
> So what can I do from there ?

Before you start randomly zeroing stuff, do you have any files in
pg_clog?  What does that directory look like?

--
Alvaro Herrera                                http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.

I have only 2 files in this dir :

total 336
-rw------- 1 postgres postgres 262144 2006-11-15 18:34 0000
-rw------- 1 postgres postgres  73728 2008-02-05 14:09 0001

I thought clog files were only files related to transactions in progress ? There's no way to tell psql to ignore all current transactions and act like if they were rolled back or something ?