Postgres Pro
Downloads
Home   >   mailing lists

Thread: invalid length of startup packet

invalid length of startup packet

From
"Dan The Man"
Date:
Hi,

I've been using postgresql v810 for a couple of weeks.  I notice when I
start it up I get 'invalid lenth of startup packet'  On a database that I
just initialized I get ' incomplete startup packet' after I first started
it.   This never happened that I can see in 8.0.3.

Does any one know if that can be safely ignored?

~DjK


< % %2005-11-28 07:31:10 EST>LOG:  database system was interrupted at
2005-11-28 03:57:12 EST
< % %2005-11-28 07:31:10 EST>LOG:  checkpoint record is at 13/89BD9728
< % %2005-11-28 07:31:10 EST>LOG:  redo record is at 13/89BD9728; undo
record is at 0/0; shutdown FALSE
< % %2005-11-28 07:31:10 EST>LOG:  next transaction ID: 81691647; next OID:
92997625
< % %2005-11-28 07:31:10 EST>LOG:  next MultiXactId: 1; next
MultiXactOffset: 0
< % %2005-11-28 07:31:10 EST>LOG:  database system was not properly shut
down; automatic recovery in progress
< % %2005-11-28 07:31:10 EST>LOG:  record with zero length at 13/89BD9778
< % %2005-11-28 07:31:10 EST>LOG:  redo is not required
< % %2005-11-28 07:31:10 EST>LOG:  database system is ready
< % %2005-11-28 07:31:10 EST>LOG:  transaction ID wrap limit is 1155432042,
limited by database "dan_test_bum"
<[unknown] %[unknown] %2005-11-28 09:03:07 EST>LOG:  invalid length of
startup packet
< % %2005-11-28 09:43:37 EST>LOG:  autovacuum: processing database
"initial_db"
< % %2005-11-28 09:53:37 EST>LOG:  autovacuum: processing data

Re: invalid length of startup packet

From
Guido Barosio
Date:
I can notice:
 
< % %2005-11-28 07:31:10 EST>LOG:  database system was not properly shut
down; automatic recovery in progress
 
At a first sight, you didn't shut down the database as expected. (pg_ctl stop -m smart/fast/immediate). Now I don't know if this is the root cause for that message, but a clean stop, and start should show you if this is related or not.
 
Just 2 cents, every little helps, best regards,
Guido Barosio

 
On 11/29/05, Dan The Man <bitsandbytes88@hotmail.com> wrote:
Hi,

I've been using postgresql v810 for a couple of weeks.  I notice when I
start it up I get 'invalid lenth of startup packet'  On a database that I
just initialized I get ' incomplete startup packet' after I first started
it.   This never happened that I can see in 8.0.3.

Does any one know if that can be safely ignored?

~DjK


< % %2005-11-28 07:31:10 EST>LOG:  database system was interrupted at
2005-11-28 03:57:12 EST
< % %2005-11-28 07:31:10 EST>LOG:  checkpoint record is at 13/89BD9728
< % %2005-11-28 07:31:10 EST>LOG:  redo record is at 13/89BD9728; undo
record is at 0/0; shutdown FALSE
< % %2005-11-28 07:31:10 EST>LOG:  next transaction ID: 81691647; next OID:
92997625
< % %2005-11-28 07:31:10 EST>LOG:  next MultiXactId: 1; next
MultiXactOffset: 0
< % %2005-11-28 07:31:10 EST>LOG:  database system was not properly shut
down; automatic recovery in progress
< % %2005-11-28 07:31:10 EST>LOG:  record with zero length at 13/89BD9778
< % %2005-11-28 07:31:10 EST>LOG:  redo is not required
< % %2005-11-28 07:31:10 EST>LOG:  database system is ready
< % %2005-11-28 07:31:10 EST>LOG:  transaction ID wrap limit is 1155432042,
limited by database "dan_test_bum"
<[unknown] %[unknown] %2005-11-28 09:03:07 EST>LOG:  invalid length of
startup packet
< % %2005-11-28 09:43:37 EST>LOG:  autovacuum: processing database
"initial_db"
< % %2005-11-28 09:53:37 EST>LOG:  autovacuum: processing data



---------------------------(end of broadcast)---------------------------
TIP 3: Have you checked our extensive FAQ?

              http://www.postgresql.org/docs/faq



--
"Adopting the position that you are smarter than an automatic
optimization algorithm is generally a good way to achieve less
performance, not more" - Tom Lane.

Re: invalid length of startup packet

From
Tom Lane
Date:
"Dan The Man" <bitsandbytes88@hotmail.com> writes:
> I've been using postgresql v810 for a couple of weeks.  I notice when I
> start it up I get 'invalid lenth of startup packet'  On a database that I
> just initialized I get ' incomplete startup packet' after I first started
> it.   This never happened that I can see in 8.0.3.

That has nothing to do with the state of the database --- it's an
indication of broken client software trying to connect, or possibly
something portscanning your system.

            regards, tom lane