Hi Rafal,
Some replies to your comments.
> I also wonder about the way you satisfy R02. What you do is to note
> master's binlog position in the backup_progress log. But the position is
> not saved in the backup image and thus will not be shown on RESTORE.
Writing it to the progress log was requested. I have updated the worklog
accordingly.
> Saving it in the backup image will be very difficult, because that
> requires changing the format of the backup image. But if we don't do
> that, does that still satisfy requirement R02?
I agree and that should be done in WL#4105. I have made a note in that
worklog referencing the work done here.
...
>> + /*
>> + If we are a connected slave, write master's binlog information to
>> + the progress log for later use.
>> + */
>> + if (obs::is_slave())
>> + info.m_ctx.report_master_binlog_pos(active_mi);
>> +
>
> This might report incorrect position. We want to capture the exact
> position at the VP time, i.e in between sch.lock() and sch.unlock()
> calls. Since we can't do anything time consuming during that time, we
> should use the same trick as for reporting binlog position:
I have made the changes accordingly and as we discussed on IRC.
...
All other items cited have been corrected.
Chuck
