From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
To: Yury Kotov <yury-kotov@yandex-team.ru>
Cc: Juan Quintela <quintela@redhat.com>, Stefan Weil <sw@weilnetz.de>,
Peter Crosthwaite <crosthwaite.peter@gmail.com>,
"open list:Overall" <qemu-devel@nongnu.org>,
yc-core@yandex-team.ru, Paolo Bonzini <pbonzini@redhat.com>,
Richard Henderson <rth@twiddle.net>
Subject: Re: [Qemu-devel] [PATCH v4 3/3] tests/migration: Add a test for auto converge
Date: Tue, 23 Jul 2019 18:34:19 +0100 [thread overview]
Message-ID: <20190723173419.GQ2719@work-vm> (raw)
In-Reply-To: <20190723134215.25095-4-yury-kotov@yandex-team.ru>
* Yury Kotov (yury-kotov@yandex-team.ru) wrote:
> Signed-off-by: Yury Kotov <yury-kotov@yandex-team.ru>
OK, I think that's worth a go; lets see how it does in heavy CI systems/
Dave
> ---
> tests/migration-test.c | 103 ++++++++++++++++++++++++++++++++++++-----
> 1 file changed, 92 insertions(+), 11 deletions(-)
>
> diff --git a/tests/migration-test.c b/tests/migration-test.c
> index a4feb9545d..b783ae47b3 100644
> --- a/tests/migration-test.c
> +++ b/tests/migration-test.c
> @@ -241,6 +241,17 @@ static int64_t read_ram_property_int(QTestState *who, const char *property)
> return result;
> }
>
> +static int64_t read_migrate_property_int(QTestState *who, const char *property)
> +{
> + QDict *rsp_return;
> + int64_t result;
> +
> + rsp_return = migrate_query(who);
> + result = qdict_get_try_int(rsp_return, property, 0);
> + qobject_unref(rsp_return);
> + return result;
> +}
> +
> static uint64_t get_migration_pass(QTestState *who)
> {
> return read_ram_property_int(who, "dirty-sync-count");
> @@ -255,20 +266,22 @@ static void read_blocktime(QTestState *who)
> qobject_unref(rsp_return);
> }
>
> +static bool check_migration_status(QTestState *who, const char *status)
> +{
> + bool completed;
> + char *current_status;
> +
> + current_status = migrate_query_status(who);
> + completed = strcmp(current_status, status) == 0;
> + g_assert_cmpstr(current_status, !=, "failed");
> + g_free(current_status);
> + return completed;
> +}
> +
> static void wait_for_migration_status(QTestState *who,
> const char *goal)
> {
> - while (true) {
> - bool completed;
> - char *status;
> -
> - status = migrate_query_status(who);
> - completed = strcmp(status, goal) == 0;
> - g_assert_cmpstr(status, !=, "failed");
> - g_free(status);
> - if (completed) {
> - return;
> - }
> + while (!check_migration_status(who, goal)) {
> usleep(1000);
> }
> }
> @@ -1121,6 +1134,73 @@ static void test_migrate_fd_proto(void)
> test_migrate_end(from, to, true);
> }
>
> +static void test_migrate_auto_converge(void)
> +{
> + char *uri = g_strdup_printf("unix:%s/migsocket", tmpfs);
> + QTestState *from, *to;
> + int64_t remaining, percentage;
> +
> + /*
> + * We want the test to be fast enough, but stable.
> + * Throttle percentages are chosen to cover all cases (init, increment, max)
> + */
> + static const int64_t expected_pcts[] = { 0, 1, 51, 98 };
> + const int64_t max_bandwidth = 200000000; /* ~200Mb/s */
> + const int64_t downtime_limit = 50; /* 50ms */
> + /*
> + * We migrate through unix-socket (> 500Mb/s).
> + * Thus, expected migration speed ~= bandwidth limit (< 500Mb/s).
> + * So, we can predict expected_threshold
> + */
> + const int64_t expected_threshold = max_bandwidth * downtime_limit / 1000;
> +
> + if (test_migrate_start(&from, &to, uri, false, false)) {
> + return;
> + }
> +
> + migrate_set_capability(from, "auto-converge", true);
> + migrate_set_parameter_int(from, "cpu-throttle-initial", expected_pcts[1]);
> + migrate_set_parameter_int(from, "cpu-throttle-increment",
> + expected_pcts[2] - expected_pcts[1]);
> + migrate_set_parameter_int(from, "max-cpu-throttle", expected_pcts[3]);
> +
> + migrate_set_parameter_int(from, "max-bandwidth", max_bandwidth);
> + migrate_set_parameter_int(from, "downtime-limit", downtime_limit);
> +
> + /* To check remaining size after precopy */
> + migrate_set_capability(from, "pause-before-switchover", true);
> +
> + /* Wait for the first serial output from the source */
> + wait_for_serial("src_serial");
> +
> + migrate(from, uri, "{}");
> +
> + /*
> + * Wait for pre-switchover status to check last throttle percentage
> + * and remaining. These values will be zeroed later
> + */
> + wait_for_migration_status(from, "pre-switchover");
> +
> + /* We expect that migration can't converge without throttling */
> + percentage = read_migrate_property_int(from, "cpu-throttle-percentage");
> + g_assert_cmpint(percentage, >, 0);
> +
> + remaining = read_ram_property_int(from, "remaining");
> + g_assert_cmpint(remaining, <, expected_threshold);
> +
> + wait_command(from, "{ 'execute': 'migrate-continue' , 'arguments':"
> + " { 'state': 'pre-switchover' } }");
> +
> + qtest_qmp_eventwait(to, "RESUME");
> +
> + wait_for_serial("dest_serial");
> + wait_for_migration_complete(from);
> +
> + g_free(uri);
> +
> + test_migrate_end(from, to, true);
> +}
> +
> int main(int argc, char **argv)
> {
> char template[] = "/tmp/migration-test-XXXXXX";
> @@ -1176,6 +1256,7 @@ int main(int argc, char **argv)
> /* qtest_add_func("/migration/ignore_shared", test_ignore_shared); */
> qtest_add_func("/migration/xbzrle/unix", test_xbzrle_unix);
> qtest_add_func("/migration/fd_proto", test_migrate_fd_proto);
> + qtest_add_func("/migration/auto_converge", test_migrate_auto_converge);
>
> ret = g_test_run();
>
> --
> 2.22.0
>
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
next prev parent reply other threads:[~2019-07-23 17:34 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-07-23 13:42 [Qemu-devel] [PATCH v4 0/3] High downtime with 95+ throttle pct Yury Kotov
2019-07-23 13:42 ` [Qemu-devel] [PATCH v4 1/3] qemu-thread: Add qemu_cond_timedwait Yury Kotov
2019-07-23 13:42 ` [Qemu-devel] [PATCH v4 2/3] cpus: Fix throttling during vm_stop Yury Kotov
2019-07-23 13:42 ` [Qemu-devel] [PATCH v4 3/3] tests/migration: Add a test for auto converge Yury Kotov
2019-07-23 17:34 ` Dr. David Alan Gilbert [this message]
2019-07-23 17:34 ` Dr. David Alan Gilbert
2019-08-07 7:42 ` [Qemu-devel] [PATCH v4 0/3] High downtime with 95+ throttle pct Yury Kotov
2019-08-15 9:13 ` Yury Kotov
2019-08-19 16:39 ` Paolo Bonzini
2019-08-19 17:11 ` Paolo Bonzini
2019-08-23 8:27 ` Yury Kotov
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190723173419.GQ2719@work-vm \
--to=dgilbert@redhat.com \
--cc=crosthwaite.peter@gmail.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=quintela@redhat.com \
--cc=rth@twiddle.net \
--cc=sw@weilnetz.de \
--cc=yc-core@yandex-team.ru \
--cc=yury-kotov@yandex-team.ru \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).