Re: Broken hint bits (freeze)

On Fri, Jun 23, 2017 at 8:18 PM, Bruce Momjian <bruce(at)momjian(dot)us> wrote:
> On Fri, Jun 23, 2017 at 08:10:17AM +0530, Amit Kapila wrote:
>> On Wed, Jun 21, 2017 at 10:03 PM, Bruce Momjian <bruce(at)momjian(dot)us> wrote:
>> > On Wed, Jun 21, 2017 at 07:49:21PM +0530, Amit Kapila wrote:
>> >> On Tue, Jun 20, 2017 at 7:24 PM, Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>> >> > Hmm. I think we need something that works with lesser effort because
>> >> > not all users will be as knowledgeable as you are, so if they make any
>> >> > mistakes in copying the file manually, it can lead to problems. How
>> >> > about issuing a notification (XLogArchiveNotifySeg) in shutdown
>> >> > checkpoint if archiving is enabled?
>> >> >
>> >>
>> >> I have thought more about the above solution and it seems risky to
>> >> notify archiver for incomplete WAL segments (which will be possible in
>> >> this case as there is no guarantee that Checkpoint record will fill
>> >> the segment). So, it seems to me we should update the document unless
>> >> you or someone has some solution to this problem.
>> >
>> > The over-arching question is how do we tell users to verify that the WAL
>> > has been replayed on the standby? I am thinking we would say that for
>> > streaming replication, the "Latest checkpoint location" should match on
>> > the primary and standby, while for log shipping, the standbys should be
>> > exactly one WAL file less than the primary.
>> >
>>
>> I am not sure if we can say "standbys should be exactly one WAL file
>> less than the primary" because checkpoint can create few more WAL
>> segments for future use. I think to make this work user needs to
>> carefully just copy the next WAL segment (next to the last file in
>> standby) which will contain checkpoint record. Ideally, there should
>> be some way either in form of a tool or a functionality in the
>> database server with which this last file can be copied but I think in
>> the absence of that we can at least document this fact.
>
> I was not clear. I was not saying there can be only one extra WAL file.
> I am saying the "Latest checkpoint location" should be one WAL file
> farther on the master. I think the big problem is that we need a full
> replay of that WAL file, not just having it one less than the master.
>

If the user has properly shutdown, then that last file should only
have checkpoint record, is it safe to proceed with upgrade without
actually copying that file?

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com