Report from Marti Raudsepp

Expand ops/sec by two digits to maintain alignment on servers with fast
I/O subsystems, e.g. can now display < 10M ops/sec with consistent
alignment.

I didn't realize that ALIGNOF_XLOG_BUFFER was used in pg_test_fsync.

The -h option was not supported by many tools, and not documented, so
remove them for consistency from pg_upgrade, pg_test_fsync, and
pg_test_timing.

The semantics of signal(2) are more variable than one could wish; in
particular, on strict-POSIX platforms the signal handler will be reset
to SIG_DFL when the signal is delivered.  This demonstrably breaks
pg_test_fsync's use of SIGALRM.  The other changes I made are not
absolutely necessary today, because the called handlers all exit the
program anyway.  But it seems like a good general practice to use
pqsignal() exclusively in Postgres code, now that we have it available
everywhere.

Clarify the docs explaining what commit_delay does, and add a
recommendation about a useful value for it, namely half of the single-page
fsync time reported by pg_test_fsync.  This is informed by testing of
the new-in-9.3 implementation of commit_delay; in prior versions it
was far harder to arrive at a useful setting.

In passing, do some wordsmithing and markup-fixing in the same general
area.

Also, change pg_test_fsync's default time-per-test from 2 seconds to 5.
The old value was about the minimum at which the results could be taken
seriously at all, and so seems a tad optimistic as a default.

Peter Geoghegan, reviewed by Noah Misch; some additional editing by me

consistency.

Per suggestion from Tom.

Per compiler warning.

e.g. fsync  2103.613 ops/sec (   475 microsecs/op)

Peter Geoghegan

commit-fest.

comments about the alarm method used on Win32.

thread.

test, rather than a number of test cycles.  Changes -o/cycles option to
-s/seconds.

Marti Raudsepp, with additional paranoia by me.

Foremost, it should go to stdout.

O_DIRECT behavior.

This isn't fully tested as yet, in particular I'm not sure that the
"foo--unpackaged--1.0.sql" scripts are OK.  But it's time to get some
buildfarm cycles on it.

sepgsql is not converted to an extension, mainly because it seems to
require a very nonstandard installation process.

Dimitri Fontaine and Tom Lane

always 8k writes, per suggestion from Tom.  Also adjust open_sync output
layout.

it is 8k as expected.

Reduce #includes to minimum actually needed; in particular include
postgres_fe.h not postgres.h, so as to stop build failures on some
platforms.

Use get_progname() instead of hardwired program name; improve error
checking for command line syntax; bring error messages into line with
style guidelines; include strerror result in die() cases.

Per buildfarm.

Un-break Windows build (I hope) by making the HAVE_FSYNC_WRITETHROUGH
code match the backend.  Fix incorrect program help message.  static-ize
all functions.

Actually rename the program, rather than just claiming we did.  Hook it
into the build system.  Get rid of useless dependency on libpq.  Clean up
#include list and messy whitespace.