Re: PQputline error with pg_restore

[Jennifer I Drake/O/VCU <drakeji@xxxxxxx>]

The postmaster stderr was being sent to /dev/null, so I changed that and was able to generate a log file.  Unfortunately, I'm not sure what the output in the log file means - I don't have a lot of experience with this sort of thing and would greatly appreciate any help.  Here's the info from the log:

postmaster successfully started
LOG:  database system was shut down at 2006-02-09 15:37:30 EST
LOG:  checkpoint record is at 1/2C118B80
LOG:  redo record is at 1/2C118B80; undo record is at 0/0; shutdown TRUE
LOG:  next transaction id: 6659; next oid: 19331162
LOG:  database system is ready
WARNING:  Attribute "piece" has an unknown type
    Proceeding with relation creation anyway
WARNING:  Attribute "piece" has an unknown type
    Proceeding with relation creation anyway
LOG:  recycled transaction log file 000000010000002B
LOG:  recycled transaction log file 000000010000002C
LOG:  recycled transaction log file 000000010000002D
LOG:  recycled transaction log file 000000010000002E
LOG:  recycled transaction log file 000000010000002F
LOG:  recycled transaction log file 0000000100000030
LOG:  recycled transaction log file 0000000100000031
LOG:  recycled transaction log file 0000000100000032
LOG:  recycled transaction log file 0000000100000033
LOG:  recycled transaction log file 0000000100000034
ERROR:  copy: line 178286, overflow on numeric ABS(value) >= 10^3 for field with precision 5 scale 3
FATAL:  Socket command type 8 unknown
LOG:  recycled transaction log file 0000000100000036
LOG:  recycled transaction log file 0000000100000037
LOG:  recycled transaction log file 0000000100000035

Thanks,
Jennifer

-----pgsql-general-owner@xxxxxxxxxxxxxx wrote: -----
> You should really update to something more current than 7.3.2 :-(
>
> The first thing to do is get more info about the error, but
> unfortunately that release of pg_restore isn't going to tell you what
> the error message from PQputline is.  So you'll have to try to find
> out from the postmaster's log.
>
> You'll need to look at the startup script Mandrake uses for postgres
> to see where it sends the postmaster's stderr, but I wouldn't be
> surprised to find that it sends to /dev/null :-(.  You can change the
> script to redirect to some real file and then restart the postmaster and
> then try the restore again.  Or reconfigure things so that the
> postmaster sends its log messages to syslog --- though this may take
> some fooling with syslog's configuration as well as with postgresql.conf.
> (Syslog is probably a better choice for production purposes --- if you
> redirect to a file, that file will continue to grow as long as the
> postmaster runs.)
>
> Once you've managed to see the underlying error message, if it doesn't
> make things clear then pass the info along and we'll try to help.
>
> BTW, another thing you could try is having pg_restore just generate
> a SQL script, and then feed the SQL script to psql.  psql will probably
> be more cooperative about showing the underlying message.
>
>             	regards, tom lane