Обсуждение: archive logs recovery

"Sriram Dandapani" <sdandapani@counterpane.com> writes:
> I have a situation where the database needs to be restarted
> periodically. Will this affect recovery of a database using a base
> online backup that took place earlier than the restart.

No.  But why do you need to shut down the database?  Most people
around here seem to want 24x7 operation ...

            regards, tom lane

I have a very high traffic db with lots of table partitions. The only
reliable way to drop the partitioned tables is to prevent any
insert/select on the parent table , otherwise, the drop table locks out
the inserts, and strangely enough, the drop table command goes into a
wait state. During shutdown and restart, I switch the pg_hba.conf to
disallow client access for a brief period when the tables are dropped.

-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Thursday, August 31, 2006 1:20 PM
To: Sriram Dandapani
Cc: pgsql-admin@postgresql.org
Subject: Re: [ADMIN] archive logs recovery

"Sriram Dandapani" <sdandapani@counterpane.com> writes:
> I have a situation where the database needs to be restarted
> periodically. Will this affect recovery of a database using a base
> online backup that took place earlier than the restart.

No.  But why do you need to shut down the database?  Most people
around here seem to want 24x7 operation ...

            regards, tom lane

"Sriram Dandapani" <sdandapani@counterpane.com> writes:
> I have a very high traffic db with lots of table partitions. The only
> reliable way to drop the partitioned tables is to prevent any
> insert/select on the parent table , otherwise, the drop table locks out
> the inserts, and strangely enough, the drop table command goes into a
> wait state. During shutdown and restart, I switch the pg_hba.conf to
> disallow client access for a brief period when the tables are dropped.

You do know you can change pg_hba.conf without a restart?

As for the drop going into a wait state, you might want to look around
for client code that's sitting idle with an open transaction.  The drop
should not block for longer than whatever is your longest transaction
touching that table.

            regards, tom lane

The issue is that the drop table immediately puts the jdbc insert into a
wait state and goes into a wait state itself.

-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Thursday, August 31, 2006 1:37 PM
To: Sriram Dandapani
Cc: pgsql-admin@postgresql.org
Subject: Re: [ADMIN] archive logs recovery

"Sriram Dandapani" <sdandapani@counterpane.com> writes:
> I have a very high traffic db with lots of table partitions. The only
> reliable way to drop the partitioned tables is to prevent any
> insert/select on the parent table , otherwise, the drop table locks
out
> the inserts, and strangely enough, the drop table command goes into a
> wait state. During shutdown and restart, I switch the pg_hba.conf to
> disallow client access for a brief period when the tables are dropped.

You do know you can change pg_hba.conf without a restart?

As for the drop going into a wait state, you might want to look around
for client code that's sitting idle with an open transaction.  The drop
should not block for longer than whatever is your longest transaction
touching that table.

            regards, tom lane

Also, the dynamic pg_hba changes work only for new connections that are
being established via jdbc. Existing connections seem to able to make
connections.

-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Thursday, August 31, 2006 1:37 PM
To: Sriram Dandapani
Cc: pgsql-admin@postgresql.org
Subject: Re: [ADMIN] archive logs recovery

"Sriram Dandapani" <sdandapani@counterpane.com> writes:
> I have a very high traffic db with lots of table partitions. The only
> reliable way to drop the partitioned tables is to prevent any
> insert/select on the parent table , otherwise, the drop table locks
out
> the inserts, and strangely enough, the drop table command goes into a
> wait state. During shutdown and restart, I switch the pg_hba.conf to
> disallow client access for a brief period when the tables are dropped.

You do know you can change pg_hba.conf without a restart?

As for the drop going into a wait state, you might want to look around
for client code that's sitting idle with an open transaction.  The drop
should not block for longer than whatever is your longest transaction
touching that table.

            regards, tom lane