[v3,07/10] Reset the auto-converge counter at every checkpoint.

From: "Rao, Lei" <lei.rao@intel.com>

if we don't reset the auto-converge counter,
it will continue to run with COLO running,
and eventually the system will hang due to the
CPU throttle reaching DEFAULT_MIGRATE_MAX_CPU_THROTTLE.

Signed-off-by: Lei Rao <lei.rao@intel.com>
---
 migration/colo.c |  4 ++++
 migration/ram.c  | 10 ++++++++++
 migration/ram.h  |  1 +
 3 files changed, 15 insertions(+)

* leirao (lei.rao@intel.com) wrote:
> From: "Rao, Lei" <lei.rao@intel.com>
> 
> if we don't reset the auto-converge counter,
> it will continue to run with COLO running,
> and eventually the system will hang due to the
> CPU throttle reaching DEFAULT_MIGRATE_MAX_CPU_THROTTLE.
> 
> Signed-off-by: Lei Rao <lei.rao@intel.com>
> ---
>  migration/colo.c |  4 ++++
>  migration/ram.c  | 10 ++++++++++
>  migration/ram.h  |  1 +
>  3 files changed, 15 insertions(+)
> 
> diff --git a/migration/colo.c b/migration/colo.c
> index 1aaf316..723ffb8 100644
> --- a/migration/colo.c
> +++ b/migration/colo.c
> @@ -459,6 +459,10 @@ static int colo_do_checkpoint_transaction(MigrationState *s,
>      if (ret < 0) {
>          goto out;
>      }
> +
> +    if (migrate_auto_converge()) {
> +        mig_throttle_counter_reset();
> +    }
>      /*
>       * Only save VM's live state, which not including device state.
>       * TODO: We may need a timeout mechanism to prevent COLO process
> diff --git a/migration/ram.c b/migration/ram.c
> index 72143da..e795a8d 100644
> --- a/migration/ram.c
> +++ b/migration/ram.c
> @@ -652,6 +652,16 @@ static void mig_throttle_guest_down(uint64_t bytes_dirty_period,
>      }
>  }
>  
> +void mig_throttle_counter_reset(void)
> +{
> +    RAMState *rs = ram_state;
> +
> +    rs->time_last_bitmap_sync = qemu_clock_get_ms(QEMU_CLOCK_REALTIME);
> +    rs->num_dirty_pages_period = 0;
> +    rs->bytes_xfer_prev = ram_counters.transferred;
> +    cpu_throttle_stop();

I think this is right, so:

Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>

however, do you really need the cpu_throttle_stop?  Shouldn't the
previous iteration have called that in migration_iteration_finish() ?

Dave

> +}
> +
>  /**
>   * xbzrle_cache_zero_page: insert a zero page in the XBZRLE cache
>   *
> diff --git a/migration/ram.h b/migration/ram.h
> index 6378bb3..3f78175 100644
> --- a/migration/ram.h
> +++ b/migration/ram.h
> @@ -50,6 +50,7 @@ bool ramblock_is_ignored(RAMBlock *block);
>  int xbzrle_cache_resize(uint64_t new_size, Error **errp);
>  uint64_t ram_bytes_remaining(void);
>  uint64_t ram_bytes_total(void);
> +void mig_throttle_counter_reset(void);
>  
>  uint64_t ram_pagesize_summary(void);
>  int ram_save_queue_pages(const char *rbname, ram_addr_t start, ram_addr_t len);
> -- 
> 1.8.3.1
>

After testing, I think you are right.
Will remove the cpu_throttle_stop() in V4.

Thanks,
Lei.

-----Original Message-----
From: Dr. David Alan Gilbert <dgilbert@redhat.com> 
Sent: Thursday, March 25, 2021 12:40 AM
To: Rao, Lei <lei.rao@intel.com>
Cc: Zhang, Chen <chen.zhang@intel.com>; lizhijian@cn.fujitsu.com; jasowang@redhat.com; quintela@redhat.com; pbonzini@redhat.com; lukasstraub2@web.de; qemu-devel@nongnu.org
Subject: Re: [PATCH v3 07/10] Reset the auto-converge counter at every checkpoint.

* leirao (lei.rao@intel.com) wrote:
> From: "Rao, Lei" <lei.rao@intel.com>
> 
> if we don't reset the auto-converge counter, it will continue to run 
> with COLO running, and eventually the system will hang due to the CPU 
> throttle reaching DEFAULT_MIGRATE_MAX_CPU_THROTTLE.
> 
> Signed-off-by: Lei Rao <lei.rao@intel.com>
> ---
>  migration/colo.c |  4 ++++
>  migration/ram.c  | 10 ++++++++++
>  migration/ram.h  |  1 +
>  3 files changed, 15 insertions(+)
> 
> diff --git a/migration/colo.c b/migration/colo.c index 
> 1aaf316..723ffb8 100644
> --- a/migration/colo.c
> +++ b/migration/colo.c
> @@ -459,6 +459,10 @@ static int colo_do_checkpoint_transaction(MigrationState *s,
>      if (ret < 0) {
>          goto out;
>      }
> +
> +    if (migrate_auto_converge()) {
> +        mig_throttle_counter_reset();
> +    }
>      /*
>       * Only save VM's live state, which not including device state.
>       * TODO: We may need a timeout mechanism to prevent COLO process 
> diff --git a/migration/ram.c b/migration/ram.c index 72143da..e795a8d 
> 100644
> --- a/migration/ram.c
> +++ b/migration/ram.c
> @@ -652,6 +652,16 @@ static void mig_throttle_guest_down(uint64_t bytes_dirty_period,
>      }
>  }
>  
> +void mig_throttle_counter_reset(void) {
> +    RAMState *rs = ram_state;
> +
> +    rs->time_last_bitmap_sync = qemu_clock_get_ms(QEMU_CLOCK_REALTIME);
> +    rs->num_dirty_pages_period = 0;
> +    rs->bytes_xfer_prev = ram_counters.transferred;
> +    cpu_throttle_stop();

I think this is right, so:

Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>

however, do you really need the cpu_throttle_stop?  Shouldn't the previous iteration have called that in migration_iteration_finish() ?

Dave

> +}
> +
>  /**
>   * xbzrle_cache_zero_page: insert a zero page in the XBZRLE cache
>   *
> diff --git a/migration/ram.h b/migration/ram.h index 6378bb3..3f78175 
> 100644
> --- a/migration/ram.h
> +++ b/migration/ram.h
> @@ -50,6 +50,7 @@ bool ramblock_is_ignored(RAMBlock *block);  int 
> xbzrle_cache_resize(uint64_t new_size, Error **errp);  uint64_t 
> ram_bytes_remaining(void);  uint64_t ram_bytes_total(void);
> +void mig_throttle_counter_reset(void);
>  
>  uint64_t ram_pagesize_summary(void);
>  int ram_save_queue_pages(const char *rbname, ram_addr_t start, 
> ram_addr_t len);
> --
> 1.8.3.1
> 
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK