Re: Intermittent hangs with 9.2 - Mailing list pgsql-performance

On 2013-09-11 07:43:35 -0500, Merlin Moncure wrote:
> > I've been seeing a strange issue with our Postgres install for about a year
> > now, and I was hoping someone might be able to help point me at the cause.
> > At what seem like fairly random intervals Postgres will become unresponsive
> > to the 3 application nodes it services. These periods tend to last for 10 -
> > 15 minutes before everything rights itself and the system goes back to
> > normal.
> >
> > During these periods the server will report a spike in the outbound
> > bandwidth (from about 1mbs to about 5mbs most recently), a huge spike in
> > context switches / interrupts (normal peaks are around 2k/8k respectively,
> > and during these periods they‘ve gone to 15k/22k), and a load average of
> > 100+. CPU usage stays relatively low, but it’s all system time reported,
> > user time goes to zero. It doesn‘t seem to be disk related since we’re
> > running with a shared_buffers setting of 24G, which will fit just about our
> > entire database into memory, and the IO transactions reported by the server,
> > as well as the disk reads reported by Postgres stay consistently low.
> >
> > We‘ve recently started tracking how long statements take to execute, and
> > we’re seeing some really odd numbers. A simple delete by primary key, for
> > example, from a table that contains about 280,000 rows, reportedly took
> > 18h59m46.900s. An update by primary key in that same table was reported as
> > 7d 17h 58m 30.415s. That table is frequently accessed, but obviously those
> > numbers don't seem reasonable at all.
> >
> > Some other changes we've made to postgresql.conf:
> >
> > synchronous_commit = off
> >
> > maintenance_work_mem = 1GB
> > wal_level = hot_standby
> > wal_buffers = 16MB
> >
> > max_wal_senders = 10
> >
> > wal_keep_segments = 5000
> >
> > checkpoint_segments = 128
> >
> > checkpoint_timeout = 30min
> >
> > checkpoint_completion_target = 0.9
> >
> > max_connections = 500
> >
> > The server is a Dell Poweredge R900 with 4 Xeon E7430 processors, 48GB of
> > RAM, running Cent OS 6.3.
> >
> > So far we‘ve tried disabling Transparent Huge Pages after I found a number
> > of resources online that indicated similar interrupt/context switch issues,
> > but it hasn’t resolve the problem. I managed to catch it happening once and
> > run a perf which showed:
> >
> > +  41.40%       48154  postmaster  0x347ba9         f 0x347ba9
> > +   9.55%       10956  postmaster  0x2dc820         f set_config_option
> > +   8.64%        9946  postmaster  0x5a3d4          f writeListPage
> > +   5.75%        6609  postmaster  0x5a2b0          f
> > ginHeapTupleFastCollect
> > +   2.68%        3084  postmaster  0x192483         f
> > build_implied_join_equality
> > +   2.61%        2990  postmaster  0x187a55         f build_paths_for_OR
> > +   1.86%        2131  postmaster  0x794aa          f get_collation_oid
> > +   1.56%        1822  postmaster  0x5a67e          f ginHeapTupleFastInsert
> > +   1.53%        1766  postmaster  0x1929bc         f
> > distribute_qual_to_rels
> > +   1.33%        1558  postmaster  0x249671         f cmp_numerics
> >
> > I‘m not sure what 0x347ba9 represents, or why it’s an address rather than a
> > method name.

Try converting it to something more meaningful with "addr2line", that
often has more sucess.

> > That's about the sum of it. Any help would be greatly appreciated and if you
> > want any more information about our setup, please feel free to ask.

> Reducing shared buffers to around 2gb will probably make the problem go away

That profile doesn't really look like one of the problem you are
referring to would look like.

Based on the profile I'd guess it's possible that you're seing problems
with GIN's "fastupdate" mechanism.
Try ALTER INDEX whatever SET (FASTUPDATE = OFF); VACUUM
whatever's_table for all gin indexes.

It's curious that set_config_option is so high in the profile... Any
chance you could recompile postgres with -fno-omit-frame-pointers in
CFLAGS? That would allow you to use perf -g. The performance price of
that usually is below 1% for postgres.

Greetings,

Andres Freund

--
 Andres Freund                     http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services