On Wed, Sep 14, 2016 at 12:13 AM, Kouhei Kaigai <kaigai@ak.jp.nec.com> wrote:
> It looks to me pg_buffercache tries to allocate more than 1GB using
> palloc(), when shared_buffers is more than 256GB.
>
> # show shared_buffers ;
> shared_buffers
> ----------------
> 280GB
> (1 row)
>
> # SELECT buffers, d.datname, coalesce(c.relname, '???')
> FROM (SELECT count(*) buffers, reldatabase, relfilenode
> FROM pg_buffercache group by reldatabase, relfilenode) b
> LEFT JOIN pg_database d ON d.oid = b.reldatabase
> LEFT JOIN pg_class c ON d.oid = (SELECT oid FROM pg_database
> WHERE datname = current_database())
> AND b.relfilenode = pg_relation_filenode(c.oid)
> ORDER BY buffers desc;
> ERROR: invalid memory alloc request size 1174405120
>
> It is a situation to use MemoryContextAllocHuge(), instead of palloc().
> Also, it may need a back patching?
I guess so. Although it's not very desirable for it to use that much
memory, I suppose if you have a terabyte of shared_buffers you
probably have 4GB of memory on top of that to show what they contain.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company