RE: [HACKERS] Vacuum analyze bug CAUGHT

Hi

Vacuum couldn't preserve consistency without locking.
I'm anxious about locking for system tables.

> 
> Hi
> 
> Okee, I have caught the vacuum analyse crash that was giving me a load of
> grief
>

[snip]
> 
> Then, in psql, I run and get the following result:
> 
> -----------
> 
> crashtest=> vacuum analyze;
> NOTICE:  Rel pg_type: TID 4/3: InsertTransactionInProgress 129915 
> - can't shrink relation
> NOTICE:  Rel pg_attribute: TID 23/5: InsertTransactionInProgress 
> 129915 - can't shrink relation
> NOTICE:  Rel pg_attribute: TID 23/6: InsertTransactionInProgress 
> 129915 - can't shrink relation
> NOTICE:  Rel pg_attribute: TID 23/7: InsertTransactionInProgress 
> 129915 - can't shrink relation
> NOTICE:  Rel pg_attribute: TID 23/8: InsertTransactionInProgress 
> 129915 - can't shrink relation
> NOTICE:  Rel pg_attribute: TID 23/9: InsertTransactionInProgress 
> 129915 - can't shrink relation
> NOTICE:  Rel pg_attribute: TID 23/10: InsertTransactionInProgress 
> 129915 - can't shrink relation
> NOTICE:  Rel pg_attribute: TID 23/11: InsertTransactionInProgress 
> 129915 - can't shrink relation
> NOTICE:  Rel pg_attribute: TID 23/12: InsertTransactionInProgress 
> 129915 - can't shrink relation
> NOTICE:  Rel pg_class: TID 3/22: InsertTransactionInProgress 
> 129915 - can't shrink relation

CREATE TABLE doesn't lock system tables till end of transaction.
It's a cause of these NOTICE messages.

Should we lock system tables till end of transaction ?

Moreover CREATE TABLE doesn't acquire any lock for pg_attribute            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
while tuples are inserted into pg_attribute.
Concurrent vacuum may corrupt pg_attribute. 

Regards.

Hiroshi Inoue
Inoue@tpf.co.jp