From 3c662643c4bb94aaf384e83fcfb53f6e4da0035e Mon Sep 17 00:00:00 2001 From: Tatsuo Ishii Date: Wed, 30 Aug 2023 10:03:31 +0900 Subject: [PATCH] Allow pgbench to exit immediately when any client is aborted. Previously when client was aborted due to some error during benchmarking, other clients continued their run until certain number of transactions specified -t was reached or the time specified by -T was expired. At the end, the results are printed with caution: "Run was aborted; the above results are incomplete" shows. New option "--exit-on-abort" allows pgbench to exit immediately in this case so that users could quickly fix the cause of the failure and try again another round of benchmarking. Author: Yugo Nagata Reviewed-by: Fabien COELHO, Tatsuo Ishii Discussion: https://postgr.es/m/flat/20230804130325.df32e60879c38c92bca64207%40sraoss.co.jp --- doc/src/sgml/ref/pgbench.sgml | 28 +++++++++++++++-- src/bin/pgbench/pgbench.c | 33 +++++++++++++++++++- src/bin/pgbench/t/001_pgbench_with_server.pl | 26 +++++++++++++++ 3 files changed, 83 insertions(+), 4 deletions(-) diff --git a/doc/src/sgml/ref/pgbench.sgml b/doc/src/sgml/ref/pgbench.sgml index 6c5c8afa6d..d3d6e3ef07 100644 --- a/doc/src/sgml/ref/pgbench.sgml +++ b/doc/src/sgml/ref/pgbench.sgml @@ -768,6 +768,24 @@ pgbench options d + + + + + Exit immediately when any client is aborted due to some error. Without + this option, even when a client is aborted, other clients could + continue their run as specified by + or option, and pgbench + will print an incomplete results in this case. + + + Note that serialization failures or deadlock failures do not abort the + client, so they are not affected by this option. + See for more information. + + + + @@ -985,7 +1003,8 @@ pgbench options d benchmark such as initial connection failures also exit with status 1. Errors during the run such as database errors or problems in the script will result in exit status 2. In the latter case, - pgbench will print partial results. + pgbench will print partial results if + option is not specified. @@ -2801,14 +2820,17 @@ statement latencies in milliseconds, failures and retries: start a connection to the database server / the socket for connecting the client to the database server has become invalid). In such cases all clients of this thread stop while other threads continue to work. + However, is specified, all of the + threads stop immediately in this case. Direct client errors. They lead to immediate exit from pgbench with the corresponding error message - only in the case of an internal pgbench - error (which are supposed to never occur...). Otherwise in the worst + in the case of an internal pgbench + error (which are supposed to never occur...) or when + is specified. Otherwise in the worst case they only lead to the abortion of the failed client while other clients continue their run (but some client errors are handled without an abortion of the client and reported separately, see below). Later in diff --git a/src/bin/pgbench/pgbench.c b/src/bin/pgbench/pgbench.c index 763c4b946a..617b2023b1 100644 --- a/src/bin/pgbench/pgbench.c +++ b/src/bin/pgbench/pgbench.c @@ -765,6 +765,8 @@ static int64 total_weight = 0; static bool verbose_errors = false; /* print verbose messages of all errors */ +static bool exit_on_abort = false; /* exit when any client is aborted */ + /* Builtin test scripts */ typedef struct BuiltinScript { @@ -911,6 +913,7 @@ usage(void) " -T, --time=NUM duration of benchmark test in seconds\n" " -v, --vacuum-all vacuum all four standard tables before tests\n" " --aggregate-interval=NUM aggregate data over NUM seconds\n" + " --exit-on-abort exit when any client is aborted\n" " --failures-detailed report the failures grouped by basic types\n" " --log-prefix=PREFIX prefix for transaction time log file\n" " (default: \"pgbench_log\")\n" @@ -6617,6 +6620,7 @@ main(int argc, char **argv) {"failures-detailed", no_argument, NULL, 13}, {"max-tries", required_argument, NULL, 14}, {"verbose-errors", no_argument, NULL, 15}, + {"exit-on-abort", no_argument, NULL, 16}, {NULL, 0, NULL, 0} }; @@ -6950,6 +6954,10 @@ main(int argc, char **argv) benchmarking_option_set = true; verbose_errors = true; break; + case 16: /* exit-on-abort */ + benchmarking_option_set = true; + exit_on_abort = true; + break; default: /* getopt_long already emitted a complaint */ pg_log_error_hint("Try \"%s --help\" for more information.", progname); @@ -7558,11 +7566,18 @@ threadRun(void *arg) advanceConnectionState(thread, st, &aggs); + /* + * If --exit-on-abort is used, the program is going to exit + * when any client is aborted. + */ + if (exit_on_abort && st->state == CSTATE_ABORTED) + goto done; /* * If advanceConnectionState changed client to finished state, * that's one fewer client that remains. */ - if (st->state == CSTATE_FINISHED || st->state == CSTATE_ABORTED) + else if (st->state == CSTATE_FINISHED || + st->state == CSTATE_ABORTED) remains--; } @@ -7595,6 +7610,22 @@ threadRun(void *arg) } done: + if (exit_on_abort) + { + /* + * Abort if any client is not finished, meaning some error occurred. + */ + for (int i = 0; i < nstate; i++) + { + if (state[i].state != CSTATE_FINISHED) + { + pg_log_error("Run was aborted due to an error in thread %d", + thread->tid); + exit(2); + } + } + } + disconnect_all(state, nstate); if (thread->logfile) diff --git a/src/bin/pgbench/t/001_pgbench_with_server.pl b/src/bin/pgbench/t/001_pgbench_with_server.pl index 142f966300..96be529d6b 100644 --- a/src/bin/pgbench/t/001_pgbench_with_server.pl +++ b/src/bin/pgbench/t/001_pgbench_with_server.pl @@ -1475,6 +1475,32 @@ SELECT pg_advisory_unlock_all(); # Clean up $node->safe_psql('postgres', 'DROP TABLE first_client_table, xy;'); +# Test --exit-on-abort +$node->safe_psql('postgres', + 'CREATE TABLE counter(i int); '. + 'INSERT INTO counter VALUES (0);' +); + +$node->pgbench( + '-t 10 -c 2 -j 2 --exit-on-abort', + 2, + [], + [ + qr{division by zero}, + qr{Run was aborted due to an error in thread} + ], + 'test --exit-on-abort', + { + '001_exit_on_abort' => q{ +update counter set i = i+1 returning i \gset +\if :i = 5 +\set y 1/0 +\endif +} + }); + +# Clean up +$node->safe_psql('postgres', 'DROP TABLE counter;'); # done $node->safe_psql('postgres', 'DROP TABLESPACE regress_pgbench_tap_1_ts');