Thread: Problem accessing database on nodes of a cluster

Problem accessing database on nodes of a cluster

From
Kashif Riaz
Date:
Hello admins,

I am having problems accessing a database on the nodes of a cluster. I am
working on a project that is supposed to distribute data on the nodes of a
beowulf cluster and then run queries in parallel on the nodes.

I am able to create and access a database on the master node. But when I
try to access the database on a slave node, I get the following error:

psql: FATAL:  database "db001", OID 1198223, has disappeared from pg_database
[1]+  Exit 1                  postmaster -p 9012 -D
/home/gayatri/postgres/data >logfile27 2>&1

I would really appreciate any help in this matter.

Thank you.

Sincerely,
Kashif Riaz


The log is given below:

********** ON THE MASTER NODE **********

gayatri@plundar:~$ postmaster -p 9012 -D /home/gayatri/postgres/data
>logfileP 2>&1 &
[1] 10114

gayatri@plundar:~$ cat logfileP
LOG:  could not create IPv6 socket: Address family not supported by protocol
LOG:  database system was interrupted at 2004-11-26 21:10:07 CST
LOG:  checkpoint record is at 0/2704D208
LOG:  redo record is at 0/2704D208; undo record is at 0/0; shutdown FALSE
LOG:  next transaction ID: 1325173; next OID: 1198222
LOG:  database system was not properly shut down; automatic recovery in
progress
LOG:  record with zero length at 0/2704D248
LOG:  redo is not required

gayatri@plundar:~$ createdb -p 9012 db001
CREATE DATABASE


********** ON THE SLAVE NODE **********

gayatri@plundar:~$ rlogin gumbo27

gayatri@gumbo27:~$  postmaster -p 9012 -D /home/gayatri/postgres/data
>logfile27 2>&1 &

gayatri@gumbo27:~$ cat logfile27
LOG:  could not create IPv6 socket: Address family not supported by protocol
LOG:  database system was interrupted at 2004-11-26 20:53:57 CST
LOG:  checkpoint record is at 0/2704709C
LOG:  redo record is at 0/2704709C; undo record is at 0/0; shutdown TRUE
LOG:  next transaction ID: 1325169; next OID: 1198222
LOG:  database system was not properly shut down; automatic recovery in
progress
LOG:  record with zero length at 0/270470DC
LOG:  redo is not required
LOG:  database system is ready

gayatri@gumbo27:~$ psql -p 9012 db001
psql: FATAL:  database "db001", OID 1198223, has disappeared from pg_database
[1]+  Exit 1                  postmaster -p 9012 -D
/home/gayatri/postgres/data >logfile27 2>&1









Re: Problem accessing database on nodes of a cluster

From
Tom Lane
Date:
Kashif Riaz <kriaz@cs.uno.edu> writes:
> psql: FATAL:  database "db001", OID 1198223, has disappeared from pg_database

Hmm, did you drop and recreate database db001?  There are some scenarios
where incoming backends can get confused in such scenarios.  Try
"VACUUM pg_database; CHECKPOINT".

            regards, tom lane