[Qemu-devel] [PATCH v2 0/2] throttle: improve command-line parameter documentation

[Qemu-devel] [PATCH v2 0/2] throttle: improve command-line parameter documentation

[Stefan Hajnoczi]

v2:
 * Fixed s/bps/iops/ copy-paste error in Patch 1 [Berto]
 * Rephrased warning about guest hangs and errors [Berto]
 * Added Patch 2 to hide the internal .max value from the monitor

Patch 1 fleshes out the documentation for I/O throttling command-line
parameters.

Patch 2 hides an internal value that was being exposed to users via the
monitor and caused confusion.

I ended up not adding QMP-style throttling.* names to the command-line
documentation because the names are very long and unlikely to be used.  I
couldn't see a nice way of adding them while still keeping the documentation
readable.

Stefan Hajnoczi (2):
  qemu-options: explain disk I/O throttling options
  throttle: make throttle_config(throttle_get_config()) symmetric

 tests/test-throttle.c |  8 ++++----
 util/throttle.c       | 16 ++++++++++++++++
 qemu-options.hx       | 24 ++++++++++++++++++++++++
 3 files changed, 44 insertions(+), 4 deletions(-)

-- 
2.9.3

[Qemu-devel] [PATCH v2 1/2] qemu-options: explain disk I/O throttling options

[Stefan Hajnoczi]

The disk I/O throttling options have been listed for a long time but
never explained on the QEMU man page.

Suggested-by: Nini Gu <ngu@redhat.com>
Cc: Alberto Garcia <berto@igalia.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 qemu-options.hx | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/qemu-options.hx b/qemu-options.hx
index 5633d39..c609b05 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -587,6 +587,30 @@ file sectors into the image file.
 conversion of plain zero writes by the OS to driver specific optimized
 zero write commands. You may even choose "unmap" if @var{discard} is set
 to "unmap" to allow a zero write to be converted to an UNMAP operation.
+@item bps=@var{b},bps_rd=@var{r},bps_wr=@var{w}
+Specify bandwidth throttling limits in bytes per second, either for all request
+types or for reads or writes only.  Small values can lead to timeouts or hangs
+inside the guest.  A safe minimum for disks is 2 MB/s.
+@item bps_max=@var{bm},bps_rd_max=@var{rm},bps_wr_max=@var{wm}
+Specify bursts in bytes per second, either for all request types or for reads
+or writes only.  Bursts allow the guest I/O to spike above the limit
+temporarily.
+@item iops=@var{i},iops_rd=@var{r},iops_wr=@var{w}
+Specify request rate limits in requests per second, either for all request
+types or for reads or writes only.
+@item iops_max=@var{bm},iops_rd_max=@var{rm},iops_wr_max=@var{wm}
+Specify bursts in requests per second, either for all request types or for reads
+or writes only.  Bursts allow the guest I/O to spike above the limit
+temporarily.
+@item iops_size=@var{is}
+Let every @var{is} bytes of a request count as a new request for iops
+throttling purposes.  Use this option to prevent guests from circumventing iops
+limits by sending fewer but larger requests.
+@item group=@var{g}
+Join a throttling quota group with given name @var{g}.  All drives that are
+members of the same group are accounted for together.  Use this option to
+prevent guests from circumventing throttling limits by using many small disks
+instead of a single larger disk.
 @end table
 
 By default, the @option{cache=writeback} mode is used. It will report data
-- 
2.9.3

[Qemu-devel] [PATCH v2 2/2] throttle: make throttle_config(throttle_get_config()) symmetric

[Stefan Hajnoczi]

Throttling has a weird property that throttle_get_config() does not
always return the same throttling settings that were given with
throttle_config().  In other words, the set and get functions aren't
symmetric.

If .max is 0 then the throttling code assigns a default value of .avg /
10 in throttle_config().  This is an implementation detail of the
throttling algorithm.  When throttle_get_config() is called the .max
value returned should still be 0.

Users are exposed to this quirk via "info block" or "query-block"
monitor commands.  This has caused confusion because it looks like a bug
when an unexpected value is reported.

This patch hides the .max value adjustment in throttle_get_config() and
updates test-throttle.c appropriately.

Reported-by: Nini Gu <ngu@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 tests/test-throttle.c |  8 ++++----
 util/throttle.c       | 16 ++++++++++++++++
 2 files changed, 20 insertions(+), 4 deletions(-)

diff --git a/tests/test-throttle.c b/tests/test-throttle.c
index 363b59a..3d6bb82 100644
--- a/tests/test-throttle.c
+++ b/tests/test-throttle.c
@@ -205,8 +205,8 @@ static void test_config_functions(void)
     orig_cfg.buckets[THROTTLE_OPS_READ].avg  = 69;
     orig_cfg.buckets[THROTTLE_OPS_WRITE].avg = 23;
 
-    orig_cfg.buckets[THROTTLE_BPS_TOTAL].max = 0; /* should be corrected */
-    orig_cfg.buckets[THROTTLE_BPS_READ].max  = 1; /* should not be corrected */
+    orig_cfg.buckets[THROTTLE_BPS_TOTAL].max = 0;
+    orig_cfg.buckets[THROTTLE_BPS_READ].max  = 1;
     orig_cfg.buckets[THROTTLE_BPS_WRITE].max = 120;
 
     orig_cfg.buckets[THROTTLE_OPS_TOTAL].max = 150;
@@ -246,8 +246,8 @@ static void test_config_functions(void)
     g_assert(final_cfg.buckets[THROTTLE_OPS_READ].avg  == 69);
     g_assert(final_cfg.buckets[THROTTLE_OPS_WRITE].avg == 23);
 
-    g_assert(final_cfg.buckets[THROTTLE_BPS_TOTAL].max == 15.3);/* fixed */
-    g_assert(final_cfg.buckets[THROTTLE_BPS_READ].max  == 1);   /* not fixed */
+    g_assert(final_cfg.buckets[THROTTLE_BPS_TOTAL].max == 0);
+    g_assert(final_cfg.buckets[THROTTLE_BPS_READ].max  == 1);
     g_assert(final_cfg.buckets[THROTTLE_BPS_WRITE].max == 120);
 
     g_assert(final_cfg.buckets[THROTTLE_OPS_TOTAL].max == 150);
diff --git a/util/throttle.c b/util/throttle.c
index 3817d9b..d87b421 100644
--- a/util/throttle.c
+++ b/util/throttle.c
@@ -380,6 +380,16 @@ static void throttle_fix_bucket(LeakyBucket *bkt)
     }
 }
 
+/* undo internal bucket parameter changes (see throttle_fix_bucket()) */
+static void throttle_unfix_bucket(LeakyBucket *bkt)
+{
+    double min = bkt->avg / 10;
+
+    if (bkt->max == min) {
+        bkt->max = 0;
+    }
+}
+
 /* take care of canceling a timer */
 static void throttle_cancel_timer(QEMUTimer *timer)
 {
@@ -420,7 +430,13 @@ void throttle_config(ThrottleState *ts,
  */
 void throttle_get_config(ThrottleState *ts, ThrottleConfig *cfg)
 {
+    int i;
+
     *cfg = ts->cfg;
+
+    for (i = 0; i < BUCKETS_COUNT; i++) {
+        throttle_unfix_bucket(&cfg->buckets[i]);
+    }
 }
 
 
-- 
2.9.3

Re: [Qemu-devel] [PATCH v2 1/2] qemu-options: explain disk I/O throttling options

[Alberto Garcia]

On Tue 28 Feb 2017 12:19:34 PM CET, Stefan Hajnoczi wrote:
> The disk I/O throttling options have been listed for a long time but
> never explained on the QEMU man page.
>
> Suggested-by: Nini Gu <ngu@redhat.com>
> Cc: Alberto Garcia <berto@igalia.com>
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Reviewed-by: Alberto Garcia <berto@igalia.com>

Berto

Re: [Qemu-devel] [PATCH v2 2/2] throttle: make throttle_config(throttle_get_config()) symmetric

[Alberto Garcia]

On Tue 28 Feb 2017 12:19:35 PM CET, Stefan Hajnoczi wrote:
> +/* undo internal bucket parameter changes (see throttle_fix_bucket()) */
> +static void throttle_unfix_bucket(LeakyBucket *bkt)
> +{
> +    double min = bkt->avg / 10;
> +
> +    if (bkt->max == min) {
> +        bkt->max = 0;
> +    }
> +}

I guess you could do the more general if (bkt->max < bkt->avg), but your
solution is also fine with me.

Reviewed-by: Alberto Garcia <berto@igalia.com>

Berto

Re: [Qemu-devel] [Qemu-block] [PATCH v2 2/2] throttle: make throttle_config(throttle_get_config()) symmetric

[Stefan Hajnoczi]

[-- Attachment #1: Type: text/plain, Size: 888 bytes --]

On Tue, Feb 28, 2017 at 01:17:55PM +0100, Alberto Garcia wrote:
> On Tue 28 Feb 2017 12:19:35 PM CET, Stefan Hajnoczi wrote:
> > +/* undo internal bucket parameter changes (see throttle_fix_bucket()) */
> > +static void throttle_unfix_bucket(LeakyBucket *bkt)
> > +{
> > +    double min = bkt->avg / 10;
> > +
> > +    if (bkt->max == min) {
> > +        bkt->max = 0;
> > +    }
> > +}
> 
> I guess you could do the more general if (bkt->max < bkt->avg), but your
> solution is also fine with me.
> 
> Reviewed-by: Alberto Garcia <berto@igalia.com>

I did that originally because I try to avoid floating-point equality.
The test case uses an invalid setting though (->max = 1, ->avg = 56)
which would be rejected if given on the command-line.

Thinking about this again, it's probably better to modify the test case
and use max < avg.

Will send a v3.

Stefan

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 455 bytes --]