mbox series

[v10,00/12] *** A Method for evaluating dirty page rate ***

Message ID 1600237327-33618-1-git-send-email-zhengchuan@huawei.com
Headers show
Series *** A Method for evaluating dirty page rate *** | expand

Message

Zheng Chuan Sept. 16, 2020, 6:21 a.m. UTC
v9 -> v10:
    rename find_page_matched as find_block_matched
    fix wrong termination condition in find_block_matched
    add review-by for patches

v8 -> v9:
    fix wrong index return of record_ramblock_hash_info
    optimize variable name according to review
    reset dirty_rate as -1
    change returns of compare_page_hash_info to bool

v7 -> v8:
    add atomic_read for dirtyrate status
    add error_report if set dirtyrate state failed
    change returns of save_ramblock_hash and record_ramblock_hash_info to bool
    alloc ramblock dirtyinfo array at one time
    add review-by for patches

v6 -> v7:
    fix minior comments and coding style by review
    add review-by for patches

v5 -> v6:
    fix coding style according to review
    use TARGET_PAGE_SIZE and TARGET_PAGE_BITS instead of self-defined macros
    return start-time and calc-time by qmp command

v4 -> v5:
    fix git apply failed due to meson-build
    add review-by for patches in v3

v3 -> v4:
    use crc32 to get hash result instead of md5
    add DirtyRateStatus to denote calculation status
    add some trace_calls to make it easier to debug
    fix some comments accroding to review

v2 -> v3:
    fix size_t compile warning
    fix codestyle checked by checkpatch.pl

v1 -> v2:
    use g_rand_new() to generate rand_buf
    move RAMBLOCK_FOREACH_MIGRATABLE into migration/ram.h
    add skip_sample_ramblock to filter sampled ramblock
    fix multi-numa vm coredump when query dirtyrate
    rename qapi interface and rename some structures and functions
    succeed to compile by appling each patch
    add test for migrating vm

Sometimes it is neccessary to evaluate dirty page rate before migration.
Users could decide whether to proceed migration based on the evaluation
in case of vm performance loss due to heavy workload.
Unlikey simulating dirtylog sync which could do harm on runnning vm,
we provide a sample-hash method to compare hash results for samping page.
In this way, it would have hardly no impact on vm performance.

Evaluate the dirtypage rate both on running and migration vm.
The VM specifications for migration are as follows:
- VM use 4-K page;
- the number of VCPU is 32;
- the total memory is 32Gigabit;
- use 'mempress' tool to pressurize VM(mempress 4096 1024);
- migration bandwidth is 1GB/s

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|                      |  running  |                  migrating                           |
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
| no mempress          |   4MB/s   |          8MB/s      (migrated success)               |
-------------------------------------------------------------------------------------------
| mempress 4096 1024   |  1060MB/s |     456MB/s ~ 1142MB/s (cpu throttle triggered)      |
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
| mempress 4096 4096   |  4114MB/s |     688MB/s ~ 4132MB/s (cpu throttle triggered)      |
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Test dirtyrate by qmp command like this:
1.  virsh qemu-monitor-command [vmname] '{"execute":"calc-dirty-rate", "arguments": {"calc-time": [sleep-time]}}'; 
2.  sleep specific time which is a bit larger than sleep-time
3.  virsh qemu-monitor-command [vmname] '{"execute":"query-dirty-rate"}'

The qmp command returns like this:
{"return":{"status":"measured","dirty-rate":374,"start-time":3718293,"calc-time":1},"id":"libvirt-15"}

Further test dirtyrate by libvirt api like this:
virsh getdirtyrate [vmname] [sleep-time]

Chuan Zheng (12):
  migration/dirtyrate: setup up query-dirtyrate framwork
  migration/dirtyrate: add DirtyRateStatus to denote calculation status
  migration/dirtyrate: Add RamblockDirtyInfo to store sampled page info
  migration/dirtyrate: Add dirtyrate statistics series functions
  migration/dirtyrate: move RAMBLOCK_FOREACH_MIGRATABLE into ram.h
  migration/dirtyrate: Record hash results for each sampled page
  migration/dirtyrate: Compare page hash results for recorded sampled
    page
  migration/dirtyrate: skip sampling ramblock with size below
    MIN_RAMBLOCK_SIZE
  migration/dirtyrate: Implement set_sample_page_period() and
    is_sample_period_valid()
  migration/dirtyrate: Implement calculate_dirtyrate() function
  migration/dirtyrate: Implement
    qmp_cal_dirty_rate()/qmp_get_dirty_rate() function
  migration/dirtyrate: Add trace_calls to make it easier to debug

 migration/dirtyrate.c  | 426 +++++++++++++++++++++++++++++++++++++++++++++++++
 migration/dirtyrate.h  |  70 ++++++++
 migration/meson.build  |   2 +-
 migration/ram.c        |  11 +-
 migration/ram.h        |  10 ++
 migration/trace-events |   8 +
 qapi/migration.json    |  67 ++++++++
 7 files changed, 583 insertions(+), 11 deletions(-)
 create mode 100644 migration/dirtyrate.c
 create mode 100644 migration/dirtyrate.h

Comments

Li Qiang Sept. 16, 2020, 2:49 p.m. UTC | #1
Chuan Zheng <zhengchuan@huawei.com> 于2020年9月16日周三 下午2:11写道:
>
> Compare page hash results for recorded sampled page.
>
> Signed-off-by: Chuan Zheng <zhengchuan@huawei.com>
> Signed-off-by: YanYing Zhuang <ann.zhuangyanying@huawei.com>
> Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>

Reviewed-by: Li Qiang <liq3ea@gmail.com>

> ---
>  migration/dirtyrate.c | 63 +++++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 63 insertions(+)
>
> diff --git a/migration/dirtyrate.c b/migration/dirtyrate.c
> index 5e6eedf..68561af 100644
> --- a/migration/dirtyrate.c
> +++ b/migration/dirtyrate.c
> @@ -177,6 +177,69 @@ out:
>      return ret;
>  }
>
> +static void calc_page_dirty_rate(struct RamblockDirtyInfo *info)
> +{
> +    uint32_t crc;
> +    int i;
> +
> +    for (i = 0; i < info->sample_pages_count; i++) {
> +        crc = get_ramblock_vfn_hash(info, info->sample_page_vfn[i]);
> +        if (crc != info->hash_result[i]) {
> +            info->sample_dirty_count++;
> +        }
> +    }
> +}
> +
> +static struct RamblockDirtyInfo *
> +find_block_matched(RAMBlock *block, int count,
> +                  struct RamblockDirtyInfo *infos)
> +{
> +    int i;
> +    struct RamblockDirtyInfo *matched;
> +
> +    for (i = 0; i < count; i++) {
> +        if (!strcmp(infos[i].idstr, qemu_ram_get_idstr(block))) {
> +            break;
> +        }
> +    }
> +
> +    if (i == count) {
> +        return NULL;
> +    }
> +
> +    if (infos[i].ramblock_addr != qemu_ram_get_host_addr(block) ||
> +        infos[i].ramblock_pages !=
> +            (qemu_ram_get_used_length(block) >> TARGET_PAGE_BITS)) {
> +        return NULL;
> +    }
> +
> +    matched = &infos[i];
> +
> +    return matched;
> +}
> +
> +static bool compare_page_hash_info(struct RamblockDirtyInfo *info,
> +                                  int block_count)
> +{
> +    struct RamblockDirtyInfo *block_dinfo = NULL;
> +    RAMBlock *block = NULL;
> +
> +    RAMBLOCK_FOREACH_MIGRATABLE(block) {
> +        block_dinfo = find_block_matched(block, block_count, info);
> +        if (block_dinfo == NULL) {
> +            continue;
> +        }
> +        calc_page_dirty_rate(block_dinfo);
> +        update_dirtyrate_stat(block_dinfo);
> +    }
> +
> +    if (DirtyStat.total_sample_count == 0) {
> +        return false;
> +    }
> +
> +    return true;
> +}
> +
>  static void calculate_dirtyrate(struct DirtyRateConfig config)
>  {
>      /* todo */
> --
> 1.8.3.1
>
Dr. David Alan Gilbert Sept. 17, 2020, 9:15 a.m. UTC | #2
* Chuan Zheng (zhengchuan@huawei.com) wrote:
> v9 -> v10:
>     rename find_page_matched as find_block_matched
>     fix wrong termination condition in find_block_matched
>     add review-by for patches
> 
> v8 -> v9:
>     fix wrong index return of record_ramblock_hash_info
>     optimize variable name according to review
>     reset dirty_rate as -1
>     change returns of compare_page_hash_info to bool
> 
> v7 -> v8:
>     add atomic_read for dirtyrate status
>     add error_report if set dirtyrate state failed
>     change returns of save_ramblock_hash and record_ramblock_hash_info to bool
>     alloc ramblock dirtyinfo array at one time
>     add review-by for patches
> 
> v6 -> v7:
>     fix minior comments and coding style by review
>     add review-by for patches
> 
> v5 -> v6:
>     fix coding style according to review
>     use TARGET_PAGE_SIZE and TARGET_PAGE_BITS instead of self-defined macros
>     return start-time and calc-time by qmp command
> 
> v4 -> v5:
>     fix git apply failed due to meson-build
>     add review-by for patches in v3
> 
> v3 -> v4:
>     use crc32 to get hash result instead of md5
>     add DirtyRateStatus to denote calculation status
>     add some trace_calls to make it easier to debug
>     fix some comments accroding to review
> 
> v2 -> v3:
>     fix size_t compile warning
>     fix codestyle checked by checkpatch.pl
> 
> v1 -> v2:
>     use g_rand_new() to generate rand_buf
>     move RAMBLOCK_FOREACH_MIGRATABLE into migration/ram.h
>     add skip_sample_ramblock to filter sampled ramblock
>     fix multi-numa vm coredump when query dirtyrate
>     rename qapi interface and rename some structures and functions
>     succeed to compile by appling each patch
>     add test for migrating vm
> 
> Sometimes it is neccessary to evaluate dirty page rate before migration.
> Users could decide whether to proceed migration based on the evaluation
> in case of vm performance loss due to heavy workload.
> Unlikey simulating dirtylog sync which could do harm on runnning vm,
> we provide a sample-hash method to compare hash results for samping page.
> In this way, it would have hardly no impact on vm performance.
> 
> Evaluate the dirtypage rate both on running and migration vm.
> The VM specifications for migration are as follows:
> - VM use 4-K page;
> - the number of VCPU is 32;
> - the total memory is 32Gigabit;
> - use 'mempress' tool to pressurize VM(mempress 4096 1024);
> - migration bandwidth is 1GB/s
> 
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> |                      |  running  |                  migrating                           |
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> | no mempress          |   4MB/s   |          8MB/s      (migrated success)               |
> -------------------------------------------------------------------------------------------
> | mempress 4096 1024   |  1060MB/s |     456MB/s ~ 1142MB/s (cpu throttle triggered)      |
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> | mempress 4096 4096   |  4114MB/s |     688MB/s ~ 4132MB/s (cpu throttle triggered)      |
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 
> Test dirtyrate by qmp command like this:
> 1.  virsh qemu-monitor-command [vmname] '{"execute":"calc-dirty-rate", "arguments": {"calc-time": [sleep-time]}}'; 
> 2.  sleep specific time which is a bit larger than sleep-time
> 3.  virsh qemu-monitor-command [vmname] '{"execute":"query-dirty-rate"}'

Thanks; it looks like we have a full set of reviews; I'll try and add
this for the next migration pull.

> The qmp command returns like this:
> {"return":{"status":"measured","dirty-rate":374,"start-time":3718293,"calc-time":1},"id":"libvirt-15"}
> 
> Further test dirtyrate by libvirt api like this:
> virsh getdirtyrate [vmname] [sleep-time]

So do you have some libvirt patches you're going to post?

Dave

> Chuan Zheng (12):
>   migration/dirtyrate: setup up query-dirtyrate framwork
>   migration/dirtyrate: add DirtyRateStatus to denote calculation status
>   migration/dirtyrate: Add RamblockDirtyInfo to store sampled page info
>   migration/dirtyrate: Add dirtyrate statistics series functions
>   migration/dirtyrate: move RAMBLOCK_FOREACH_MIGRATABLE into ram.h
>   migration/dirtyrate: Record hash results for each sampled page
>   migration/dirtyrate: Compare page hash results for recorded sampled
>     page
>   migration/dirtyrate: skip sampling ramblock with size below
>     MIN_RAMBLOCK_SIZE
>   migration/dirtyrate: Implement set_sample_page_period() and
>     is_sample_period_valid()
>   migration/dirtyrate: Implement calculate_dirtyrate() function
>   migration/dirtyrate: Implement
>     qmp_cal_dirty_rate()/qmp_get_dirty_rate() function
>   migration/dirtyrate: Add trace_calls to make it easier to debug
> 
>  migration/dirtyrate.c  | 426 +++++++++++++++++++++++++++++++++++++++++++++++++
>  migration/dirtyrate.h  |  70 ++++++++
>  migration/meson.build  |   2 +-
>  migration/ram.c        |  11 +-
>  migration/ram.h        |  10 ++
>  migration/trace-events |   8 +
>  qapi/migration.json    |  67 ++++++++
>  7 files changed, 583 insertions(+), 11 deletions(-)
>  create mode 100644 migration/dirtyrate.c
>  create mode 100644 migration/dirtyrate.h
> 
> -- 
> 1.8.3.1
>
Zheng Chuan Sept. 17, 2020, 9:22 a.m. UTC | #3
On 2020/9/17 17:15, Dr. David Alan Gilbert wrote:
> * Chuan Zheng (zhengchuan@huawei.com) wrote:

>> v9 -> v10:

>>     rename find_page_matched as find_block_matched

>>     fix wrong termination condition in find_block_matched

>>     add review-by for patches

>>

>> v8 -> v9:

>>     fix wrong index return of record_ramblock_hash_info

>>     optimize variable name according to review

>>     reset dirty_rate as -1

>>     change returns of compare_page_hash_info to bool

>>

>> v7 -> v8:

>>     add atomic_read for dirtyrate status

>>     add error_report if set dirtyrate state failed

>>     change returns of save_ramblock_hash and record_ramblock_hash_info to bool

>>     alloc ramblock dirtyinfo array at one time

>>     add review-by for patches

>>

>> v6 -> v7:

>>     fix minior comments and coding style by review

>>     add review-by for patches

>>

>> v5 -> v6:

>>     fix coding style according to review

>>     use TARGET_PAGE_SIZE and TARGET_PAGE_BITS instead of self-defined macros

>>     return start-time and calc-time by qmp command

>>

>> v4 -> v5:

>>     fix git apply failed due to meson-build

>>     add review-by for patches in v3

>>

>> v3 -> v4:

>>     use crc32 to get hash result instead of md5

>>     add DirtyRateStatus to denote calculation status

>>     add some trace_calls to make it easier to debug

>>     fix some comments accroding to review

>>

>> v2 -> v3:

>>     fix size_t compile warning

>>     fix codestyle checked by checkpatch.pl

>>

>> v1 -> v2:

>>     use g_rand_new() to generate rand_buf

>>     move RAMBLOCK_FOREACH_MIGRATABLE into migration/ram.h

>>     add skip_sample_ramblock to filter sampled ramblock

>>     fix multi-numa vm coredump when query dirtyrate

>>     rename qapi interface and rename some structures and functions

>>     succeed to compile by appling each patch

>>     add test for migrating vm

>>

>> Sometimes it is neccessary to evaluate dirty page rate before migration.

>> Users could decide whether to proceed migration based on the evaluation

>> in case of vm performance loss due to heavy workload.

>> Unlikey simulating dirtylog sync which could do harm on runnning vm,

>> we provide a sample-hash method to compare hash results for samping page.

>> In this way, it would have hardly no impact on vm performance.

>>

>> Evaluate the dirtypage rate both on running and migration vm.

>> The VM specifications for migration are as follows:

>> - VM use 4-K page;

>> - the number of VCPU is 32;

>> - the total memory is 32Gigabit;

>> - use 'mempress' tool to pressurize VM(mempress 4096 1024);

>> - migration bandwidth is 1GB/s

>>

>> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

>> |                      |  running  |                  migrating                           |

>> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

>> | no mempress          |   4MB/s   |          8MB/s      (migrated success)               |

>> -------------------------------------------------------------------------------------------

>> | mempress 4096 1024   |  1060MB/s |     456MB/s ~ 1142MB/s (cpu throttle triggered)      |

>> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

>> | mempress 4096 4096   |  4114MB/s |     688MB/s ~ 4132MB/s (cpu throttle triggered)      |

>> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

>>

>> Test dirtyrate by qmp command like this:

>> 1.  virsh qemu-monitor-command [vmname] '{"execute":"calc-dirty-rate", "arguments": {"calc-time": [sleep-time]}}'; 

>> 2.  sleep specific time which is a bit larger than sleep-time

>> 3.  virsh qemu-monitor-command [vmname] '{"execute":"query-dirty-rate"}'

> 

> Thanks; it looks like we have a full set of reviews; I'll try and add

> this for the next migration pull.

> 

>> The qmp command returns like this:

>> {"return":{"status":"measured","dirty-rate":374,"start-time":3718293,"calc-time":1},"id":"libvirt-15"}

>>

>> Further test dirtyrate by libvirt api like this:

>> virsh getdirtyrate [vmname] [sleep-time]

> 

> So do you have some libvirt patches you're going to post?

> 

> Dave

> 


Hi, Dave.

Yes, libvirt(including libvirt-python) interface will come after this qemu series patch:)

>> Chuan Zheng (12):

>>   migration/dirtyrate: setup up query-dirtyrate framwork

>>   migration/dirtyrate: add DirtyRateStatus to denote calculation status

>>   migration/dirtyrate: Add RamblockDirtyInfo to store sampled page info

>>   migration/dirtyrate: Add dirtyrate statistics series functions

>>   migration/dirtyrate: move RAMBLOCK_FOREACH_MIGRATABLE into ram.h

>>   migration/dirtyrate: Record hash results for each sampled page

>>   migration/dirtyrate: Compare page hash results for recorded sampled

>>     page

>>   migration/dirtyrate: skip sampling ramblock with size below

>>     MIN_RAMBLOCK_SIZE

>>   migration/dirtyrate: Implement set_sample_page_period() and

>>     is_sample_period_valid()

>>   migration/dirtyrate: Implement calculate_dirtyrate() function

>>   migration/dirtyrate: Implement

>>     qmp_cal_dirty_rate()/qmp_get_dirty_rate() function

>>   migration/dirtyrate: Add trace_calls to make it easier to debug

>>

>>  migration/dirtyrate.c  | 426 +++++++++++++++++++++++++++++++++++++++++++++++++

>>  migration/dirtyrate.h  |  70 ++++++++

>>  migration/meson.build  |   2 +-

>>  migration/ram.c        |  11 +-

>>  migration/ram.h        |  10 ++

>>  migration/trace-events |   8 +

>>  qapi/migration.json    |  67 ++++++++

>>  7 files changed, 583 insertions(+), 11 deletions(-)

>>  create mode 100644 migration/dirtyrate.c

>>  create mode 100644 migration/dirtyrate.h

>>

>> -- 

>> 1.8.3.1

>>
Dr. David Alan Gilbert Sept. 17, 2020, 9:38 a.m. UTC | #4
* Zheng Chuan (zhengchuan@huawei.com) wrote:
> 

> 

> On 2020/9/17 17:15, Dr. David Alan Gilbert wrote:

> > * Chuan Zheng (zhengchuan@huawei.com) wrote:

> >> v9 -> v10:

> >>     rename find_page_matched as find_block_matched

> >>     fix wrong termination condition in find_block_matched

> >>     add review-by for patches

> >>

> >> v8 -> v9:

> >>     fix wrong index return of record_ramblock_hash_info

> >>     optimize variable name according to review

> >>     reset dirty_rate as -1

> >>     change returns of compare_page_hash_info to bool

> >>

> >> v7 -> v8:

> >>     add atomic_read for dirtyrate status

> >>     add error_report if set dirtyrate state failed

> >>     change returns of save_ramblock_hash and record_ramblock_hash_info to bool

> >>     alloc ramblock dirtyinfo array at one time

> >>     add review-by for patches

> >>

> >> v6 -> v7:

> >>     fix minior comments and coding style by review

> >>     add review-by for patches

> >>

> >> v5 -> v6:

> >>     fix coding style according to review

> >>     use TARGET_PAGE_SIZE and TARGET_PAGE_BITS instead of self-defined macros

> >>     return start-time and calc-time by qmp command

> >>

> >> v4 -> v5:

> >>     fix git apply failed due to meson-build

> >>     add review-by for patches in v3

> >>

> >> v3 -> v4:

> >>     use crc32 to get hash result instead of md5

> >>     add DirtyRateStatus to denote calculation status

> >>     add some trace_calls to make it easier to debug

> >>     fix some comments accroding to review

> >>

> >> v2 -> v3:

> >>     fix size_t compile warning

> >>     fix codestyle checked by checkpatch.pl

> >>

> >> v1 -> v2:

> >>     use g_rand_new() to generate rand_buf

> >>     move RAMBLOCK_FOREACH_MIGRATABLE into migration/ram.h

> >>     add skip_sample_ramblock to filter sampled ramblock

> >>     fix multi-numa vm coredump when query dirtyrate

> >>     rename qapi interface and rename some structures and functions

> >>     succeed to compile by appling each patch

> >>     add test for migrating vm

> >>

> >> Sometimes it is neccessary to evaluate dirty page rate before migration.

> >> Users could decide whether to proceed migration based on the evaluation

> >> in case of vm performance loss due to heavy workload.

> >> Unlikey simulating dirtylog sync which could do harm on runnning vm,

> >> we provide a sample-hash method to compare hash results for samping page.

> >> In this way, it would have hardly no impact on vm performance.

> >>

> >> Evaluate the dirtypage rate both on running and migration vm.

> >> The VM specifications for migration are as follows:

> >> - VM use 4-K page;

> >> - the number of VCPU is 32;

> >> - the total memory is 32Gigabit;

> >> - use 'mempress' tool to pressurize VM(mempress 4096 1024);

> >> - migration bandwidth is 1GB/s

> >>

> >> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

> >> |                      |  running  |                  migrating                           |

> >> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

> >> | no mempress          |   4MB/s   |          8MB/s      (migrated success)               |

> >> -------------------------------------------------------------------------------------------

> >> | mempress 4096 1024   |  1060MB/s |     456MB/s ~ 1142MB/s (cpu throttle triggered)      |

> >> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

> >> | mempress 4096 4096   |  4114MB/s |     688MB/s ~ 4132MB/s (cpu throttle triggered)      |

> >> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

> >>

> >> Test dirtyrate by qmp command like this:

> >> 1.  virsh qemu-monitor-command [vmname] '{"execute":"calc-dirty-rate", "arguments": {"calc-time": [sleep-time]}}'; 

> >> 2.  sleep specific time which is a bit larger than sleep-time

> >> 3.  virsh qemu-monitor-command [vmname] '{"execute":"query-dirty-rate"}'

> > 

> > Thanks; it looks like we have a full set of reviews; I'll try and add

> > this for the next migration pull.

> > 

> >> The qmp command returns like this:

> >> {"return":{"status":"measured","dirty-rate":374,"start-time":3718293,"calc-time":1},"id":"libvirt-15"}

> >>

> >> Further test dirtyrate by libvirt api like this:

> >> virsh getdirtyrate [vmname] [sleep-time]

> > 

> > So do you have some libvirt patches you're going to post?

> > 

> > Dave

> > 

> 

> Hi, Dave.

> 

> Yes, libvirt(including libvirt-python) interface will come after this qemu series patch:)


OK, great; please cc in jdenemar@redhat.com when you do.

Dave

> >> Chuan Zheng (12):

> >>   migration/dirtyrate: setup up query-dirtyrate framwork

> >>   migration/dirtyrate: add DirtyRateStatus to denote calculation status

> >>   migration/dirtyrate: Add RamblockDirtyInfo to store sampled page info

> >>   migration/dirtyrate: Add dirtyrate statistics series functions

> >>   migration/dirtyrate: move RAMBLOCK_FOREACH_MIGRATABLE into ram.h

> >>   migration/dirtyrate: Record hash results for each sampled page

> >>   migration/dirtyrate: Compare page hash results for recorded sampled

> >>     page

> >>   migration/dirtyrate: skip sampling ramblock with size below

> >>     MIN_RAMBLOCK_SIZE

> >>   migration/dirtyrate: Implement set_sample_page_period() and

> >>     is_sample_period_valid()

> >>   migration/dirtyrate: Implement calculate_dirtyrate() function

> >>   migration/dirtyrate: Implement

> >>     qmp_cal_dirty_rate()/qmp_get_dirty_rate() function

> >>   migration/dirtyrate: Add trace_calls to make it easier to debug

> >>

> >>  migration/dirtyrate.c  | 426 +++++++++++++++++++++++++++++++++++++++++++++++++

> >>  migration/dirtyrate.h  |  70 ++++++++

> >>  migration/meson.build  |   2 +-

> >>  migration/ram.c        |  11 +-

> >>  migration/ram.h        |  10 ++

> >>  migration/trace-events |   8 +

> >>  qapi/migration.json    |  67 ++++++++

> >>  7 files changed, 583 insertions(+), 11 deletions(-)

> >>  create mode 100644 migration/dirtyrate.c

> >>  create mode 100644 migration/dirtyrate.h

> >>

> >> -- 

> >> 1.8.3.1

> >>

> 

-- 
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
Zheng Chuan Sept. 17, 2020, 10:21 a.m. UTC | #5
On 2020/9/17 17:38, Dr. David Alan Gilbert wrote:
> * Zheng Chuan (zhengchuan@huawei.com) wrote:

>>

>>

>> On 2020/9/17 17:15, Dr. David Alan Gilbert wrote:

>>> * Chuan Zheng (zhengchuan@huawei.com) wrote:

>>>> v9 -> v10:

>>>>     rename find_page_matched as find_block_matched

>>>>     fix wrong termination condition in find_block_matched

>>>>     add review-by for patches

>>>>

>>>> v8 -> v9:

>>>>     fix wrong index return of record_ramblock_hash_info

>>>>     optimize variable name according to review

>>>>     reset dirty_rate as -1

>>>>     change returns of compare_page_hash_info to bool

>>>>

>>>> v7 -> v8:

>>>>     add atomic_read for dirtyrate status

>>>>     add error_report if set dirtyrate state failed

>>>>     change returns of save_ramblock_hash and record_ramblock_hash_info to bool

>>>>     alloc ramblock dirtyinfo array at one time

>>>>     add review-by for patches

>>>>

>>>> v6 -> v7:

>>>>     fix minior comments and coding style by review

>>>>     add review-by for patches

>>>>

>>>> v5 -> v6:

>>>>     fix coding style according to review

>>>>     use TARGET_PAGE_SIZE and TARGET_PAGE_BITS instead of self-defined macros

>>>>     return start-time and calc-time by qmp command

>>>>

>>>> v4 -> v5:

>>>>     fix git apply failed due to meson-build

>>>>     add review-by for patches in v3

>>>>

>>>> v3 -> v4:

>>>>     use crc32 to get hash result instead of md5

>>>>     add DirtyRateStatus to denote calculation status

>>>>     add some trace_calls to make it easier to debug

>>>>     fix some comments accroding to review

>>>>

>>>> v2 -> v3:

>>>>     fix size_t compile warning

>>>>     fix codestyle checked by checkpatch.pl

>>>>

>>>> v1 -> v2:

>>>>     use g_rand_new() to generate rand_buf

>>>>     move RAMBLOCK_FOREACH_MIGRATABLE into migration/ram.h

>>>>     add skip_sample_ramblock to filter sampled ramblock

>>>>     fix multi-numa vm coredump when query dirtyrate

>>>>     rename qapi interface and rename some structures and functions

>>>>     succeed to compile by appling each patch

>>>>     add test for migrating vm

>>>>

>>>> Sometimes it is neccessary to evaluate dirty page rate before migration.

>>>> Users could decide whether to proceed migration based on the evaluation

>>>> in case of vm performance loss due to heavy workload.

>>>> Unlikey simulating dirtylog sync which could do harm on runnning vm,

>>>> we provide a sample-hash method to compare hash results for samping page.

>>>> In this way, it would have hardly no impact on vm performance.

>>>>

>>>> Evaluate the dirtypage rate both on running and migration vm.

>>>> The VM specifications for migration are as follows:

>>>> - VM use 4-K page;

>>>> - the number of VCPU is 32;

>>>> - the total memory is 32Gigabit;

>>>> - use 'mempress' tool to pressurize VM(mempress 4096 1024);

>>>> - migration bandwidth is 1GB/s

>>>>

>>>> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

>>>> |                      |  running  |                  migrating                           |

>>>> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

>>>> | no mempress          |   4MB/s   |          8MB/s      (migrated success)               |

>>>> -------------------------------------------------------------------------------------------

>>>> | mempress 4096 1024   |  1060MB/s |     456MB/s ~ 1142MB/s (cpu throttle triggered)      |

>>>> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

>>>> | mempress 4096 4096   |  4114MB/s |     688MB/s ~ 4132MB/s (cpu throttle triggered)      |

>>>> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

>>>>

>>>> Test dirtyrate by qmp command like this:

>>>> 1.  virsh qemu-monitor-command [vmname] '{"execute":"calc-dirty-rate", "arguments": {"calc-time": [sleep-time]}}'; 

>>>> 2.  sleep specific time which is a bit larger than sleep-time

>>>> 3.  virsh qemu-monitor-command [vmname] '{"execute":"query-dirty-rate"}'

>>>

>>> Thanks; it looks like we have a full set of reviews; I'll try and add

>>> this for the next migration pull.

>>>

>>>> The qmp command returns like this:

>>>> {"return":{"status":"measured","dirty-rate":374,"start-time":3718293,"calc-time":1},"id":"libvirt-15"}

>>>>

>>>> Further test dirtyrate by libvirt api like this:

>>>> virsh getdirtyrate [vmname] [sleep-time]

>>>

>>> So do you have some libvirt patches you're going to post?

>>>

>>> Dave

>>>

>>

>> Hi, Dave.

>>

>> Yes, libvirt(including libvirt-python) interface will come after this qemu series patch:)

> 

> OK, great; please cc in jdenemar@redhat.com when you do.

> 

> Dave

> 

Sure, will post it asap:)

>>>> Chuan Zheng (12):

>>>>   migration/dirtyrate: setup up query-dirtyrate framwork

>>>>   migration/dirtyrate: add DirtyRateStatus to denote calculation status

>>>>   migration/dirtyrate: Add RamblockDirtyInfo to store sampled page info

>>>>   migration/dirtyrate: Add dirtyrate statistics series functions

>>>>   migration/dirtyrate: move RAMBLOCK_FOREACH_MIGRATABLE into ram.h

>>>>   migration/dirtyrate: Record hash results for each sampled page

>>>>   migration/dirtyrate: Compare page hash results for recorded sampled

>>>>     page

>>>>   migration/dirtyrate: skip sampling ramblock with size below

>>>>     MIN_RAMBLOCK_SIZE

>>>>   migration/dirtyrate: Implement set_sample_page_period() and

>>>>     is_sample_period_valid()

>>>>   migration/dirtyrate: Implement calculate_dirtyrate() function

>>>>   migration/dirtyrate: Implement

>>>>     qmp_cal_dirty_rate()/qmp_get_dirty_rate() function

>>>>   migration/dirtyrate: Add trace_calls to make it easier to debug

>>>>

>>>>  migration/dirtyrate.c  | 426 +++++++++++++++++++++++++++++++++++++++++++++++++

>>>>  migration/dirtyrate.h  |  70 ++++++++

>>>>  migration/meson.build  |   2 +-

>>>>  migration/ram.c        |  11 +-

>>>>  migration/ram.h        |  10 ++

>>>>  migration/trace-events |   8 +

>>>>  qapi/migration.json    |  67 ++++++++

>>>>  7 files changed, 583 insertions(+), 11 deletions(-)

>>>>  create mode 100644 migration/dirtyrate.c

>>>>  create mode 100644 migration/dirtyrate.h

>>>>

>>>> -- 

>>>> 1.8.3.1

>>>>

>>
Dr. David Alan Gilbert Sept. 23, 2020, 6:04 p.m. UTC | #6
Queued

* Chuan Zheng (zhengchuan@huawei.com) wrote:
> v9 -> v10:
>     rename find_page_matched as find_block_matched
>     fix wrong termination condition in find_block_matched
>     add review-by for patches
> 
> v8 -> v9:
>     fix wrong index return of record_ramblock_hash_info
>     optimize variable name according to review
>     reset dirty_rate as -1
>     change returns of compare_page_hash_info to bool
> 
> v7 -> v8:
>     add atomic_read for dirtyrate status
>     add error_report if set dirtyrate state failed
>     change returns of save_ramblock_hash and record_ramblock_hash_info to bool
>     alloc ramblock dirtyinfo array at one time
>     add review-by for patches
> 
> v6 -> v7:
>     fix minior comments and coding style by review
>     add review-by for patches
> 
> v5 -> v6:
>     fix coding style according to review
>     use TARGET_PAGE_SIZE and TARGET_PAGE_BITS instead of self-defined macros
>     return start-time and calc-time by qmp command
> 
> v4 -> v5:
>     fix git apply failed due to meson-build
>     add review-by for patches in v3
> 
> v3 -> v4:
>     use crc32 to get hash result instead of md5
>     add DirtyRateStatus to denote calculation status
>     add some trace_calls to make it easier to debug
>     fix some comments accroding to review
> 
> v2 -> v3:
>     fix size_t compile warning
>     fix codestyle checked by checkpatch.pl
> 
> v1 -> v2:
>     use g_rand_new() to generate rand_buf
>     move RAMBLOCK_FOREACH_MIGRATABLE into migration/ram.h
>     add skip_sample_ramblock to filter sampled ramblock
>     fix multi-numa vm coredump when query dirtyrate
>     rename qapi interface and rename some structures and functions
>     succeed to compile by appling each patch
>     add test for migrating vm
> 
> Sometimes it is neccessary to evaluate dirty page rate before migration.
> Users could decide whether to proceed migration based on the evaluation
> in case of vm performance loss due to heavy workload.
> Unlikey simulating dirtylog sync which could do harm on runnning vm,
> we provide a sample-hash method to compare hash results for samping page.
> In this way, it would have hardly no impact on vm performance.
> 
> Evaluate the dirtypage rate both on running and migration vm.
> The VM specifications for migration are as follows:
> - VM use 4-K page;
> - the number of VCPU is 32;
> - the total memory is 32Gigabit;
> - use 'mempress' tool to pressurize VM(mempress 4096 1024);
> - migration bandwidth is 1GB/s
> 
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> |                      |  running  |                  migrating                           |
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> | no mempress          |   4MB/s   |          8MB/s      (migrated success)               |
> -------------------------------------------------------------------------------------------
> | mempress 4096 1024   |  1060MB/s |     456MB/s ~ 1142MB/s (cpu throttle triggered)      |
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> | mempress 4096 4096   |  4114MB/s |     688MB/s ~ 4132MB/s (cpu throttle triggered)      |
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 
> Test dirtyrate by qmp command like this:
> 1.  virsh qemu-monitor-command [vmname] '{"execute":"calc-dirty-rate", "arguments": {"calc-time": [sleep-time]}}'; 
> 2.  sleep specific time which is a bit larger than sleep-time
> 3.  virsh qemu-monitor-command [vmname] '{"execute":"query-dirty-rate"}'
> 
> The qmp command returns like this:
> {"return":{"status":"measured","dirty-rate":374,"start-time":3718293,"calc-time":1},"id":"libvirt-15"}
> 
> Further test dirtyrate by libvirt api like this:
> virsh getdirtyrate [vmname] [sleep-time]
> 
> Chuan Zheng (12):
>   migration/dirtyrate: setup up query-dirtyrate framwork
>   migration/dirtyrate: add DirtyRateStatus to denote calculation status
>   migration/dirtyrate: Add RamblockDirtyInfo to store sampled page info
>   migration/dirtyrate: Add dirtyrate statistics series functions
>   migration/dirtyrate: move RAMBLOCK_FOREACH_MIGRATABLE into ram.h
>   migration/dirtyrate: Record hash results for each sampled page
>   migration/dirtyrate: Compare page hash results for recorded sampled
>     page
>   migration/dirtyrate: skip sampling ramblock with size below
>     MIN_RAMBLOCK_SIZE
>   migration/dirtyrate: Implement set_sample_page_period() and
>     is_sample_period_valid()
>   migration/dirtyrate: Implement calculate_dirtyrate() function
>   migration/dirtyrate: Implement
>     qmp_cal_dirty_rate()/qmp_get_dirty_rate() function
>   migration/dirtyrate: Add trace_calls to make it easier to debug
> 
>  migration/dirtyrate.c  | 426 +++++++++++++++++++++++++++++++++++++++++++++++++
>  migration/dirtyrate.h  |  70 ++++++++
>  migration/meson.build  |   2 +-
>  migration/ram.c        |  11 +-
>  migration/ram.h        |  10 ++
>  migration/trace-events |   8 +
>  qapi/migration.json    |  67 ++++++++
>  7 files changed, 583 insertions(+), 11 deletions(-)
>  create mode 100644 migration/dirtyrate.c
>  create mode 100644 migration/dirtyrate.h
> 
> -- 
> 1.8.3.1
> 
>
Eric Blake Sept. 23, 2020, 6:18 p.m. UTC | #7
On 9/23/20 1:04 PM, Dr. David Alan Gilbert wrote:
> Queued

> 

> * Chuan Zheng (zhengchuan@huawei.com) wrote:

>> v9 -> v10:

>>      rename find_page_matched as find_block_matched

>>      fix wrong termination condition in find_block_matched

>>      add review-by for patches


Sorry for a late review, but I suggested a QAPI tweak to 11/12.

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3226
Virtualization:  qemu.org | libvirt.org