Re: [HACKERS] Postgres Speed or lack thereof - Mailing list pgsql-hackers

From Bruce Momjian
Subject Re: [HACKERS] Postgres Speed or lack thereof
Date
Msg-id 199901200622.BAA24377@candle.pha.pa.us
Whole thread Raw
In response to Re: [HACKERS] Postgres Speed or lack thereof  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Performance test with -F (Was Postgres Speed or lack thereof)
List pgsql-hackers
> John Holland <jholland@isr.umd.edu> writes:
> > can you explain the -F flag? when is it passed to what?
> 
> -F is a command-line flag passed to the backend at backend startup.
> Since backends are normally started by the postmaster, what you
> really do in practice is to start the postmaster with "-o -F".
> For example my postmaster start script looks like
> 
> nohup postmaster -i -o "-F" >server.log 2>&1 </dev/null &
> 
> What the -F switch actually does is to disable calls to fsync(2),
> thereby allowing modified file blocks to hang around in kernel
> memory for a little while (up to 30 seconds in most Unixes)
> rather than being force-written to disk as soon as each transaction
> commits.  If the same block of the database file gets modified
> again within that time window (very likely under a repeated-update
> load), you just saved a disk write.  On the other hand, if your OS
> crashes or your power goes out in those 30 sec, you just lost a
> database update that you thought you had committed.
> 
> I'm not sure I believe the argument that omitting -F buys very much
> safety, even if you do not trust your power company.  Murphy's law
> says that a power flicker will happen in the middle of committing
> a transaction, not during the 30-second-max window between when you
> could've had the data flushed to disk if only you'd used fsync()
> and when the swapper process will fsync it on its own.  And in that
> case you have a corrupted database anyway.  So my theory is you use
> a reliable OS, and get yourself a UPS if your power company isn't
> reliable (lord knows mine ain't), and back up your database as often
> as you can.  -F buys enough speed that it's worth the small extra risk.
> 
> There are competent experts with conflicting opinions, however ;-)
> 
> See doc/README.fsync for more about -F and the implications of
> using it.

Well said, and I think we have to address this shortcoming.  Vadim has
been reluctant to turn off fsync by default.  I have been trying to come
up with some soft-fsync on my own but haven't hit on anything he agrees
with yet.

--  Bruce Momjian                        |  http://www.op.net/~candle maillist@candle.pha.pa.us            |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026
 


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: [HACKERS] Postgres Speed or lack thereof
Next
From: "Hiroshi Inoue"
Date:
Subject: SPI_prepare() doesn't work well ?