Peter Xu <pet...@redhat.com> wrote:
> On Wed, Oct 04, 2017 at 12:39:28PM +0200, Juan Quintela wrote:
>
> [...]
>
> (having some unit tests for migration is really good...)
>
>> +static void test_precopy(const char *uri)
>> +{
>> + QTestState *from, *to;
>> +
>> + test_migrate_start(&from, &to, uri);
>> +
>> + /* We want to pick a speed slow enough that the test completes
>> + * quickly, but that it doesn't complete precopy even on a slow
>> + * machine, so also set the downtime.
>> + */
>> + /* 100 ms */
>> + migrate_set_parameter(from, "downtime-limit", "100");
>> + /* 1MB/s slow*/
>
> This is 100MB/s?
>
>> + migrate_set_parameter(from, "max-bandwidth", "100000000");
>> +
>> + /* Wait for the first serial output from the source */
>> + wait_for_serial("src_serial");
>> +
>> + migrate(from, uri);
>> +
>> + wait_for_migration_pass(from);
>> +
>> + /* 1GB/s now it should converge */
>> + migrate_set_parameter(from, "max-bandwidth", "1000000000");
>
> Curious: would it possible that 1GB/s won't suffice in some cases?
> Is there any reason behind this 1GB/s value?
No.
My new test setup just uses 1GB/s
and plays with the downtime (1ms) while I want to wait for one full
cycle
and then 300ms.
Notice that the guest is around 100MB of size, so ....
>
> (I agree with Thomas that it would be good to work upon the existing
> postcopy-test.c file, at least easier to review :)
Done.
did the other way around:
rename postcopy-test.c to migrate-test.c
and then went from there.
Later, Juan.
