Обсуждение: pg_upgrade and wraparound

On 06/09/2018 03:46 AM, Alexander Shutyaev wrote:
> Hello!
> 
> I've been trying to upgrade a postgresql cluster from 9.6 to 10. I've 
> executed the pg_upgrade with the following options:
> 
>   /usr/lib/postgresql/10/bin/pg_upgrade -b /usr/lib/postgresql/9.6/bin/ 
> -B /usr/lib/postgresql/10/bin/ -d /var/lib/postgresql/9.6/main -D 
> /var/lib/postgresql/10/main -o ' -c 
> config_file=/etc/postgresql/9.6/main/postgresql.conf' -O ' -c 
> config_file=/etc/postgresql/10/main/postgresql.conf'
> 
> The upgrade operation failed after several hours with the following error:
> 
> database is not accepting commands to avoid wraparound data loss in 
> database with OID 0

Do you know which database has an OID of 0?

> 
> Earlier in the log there are a lot of messages like
> 
> pg_restore: executing BLOB 1740736966
> pg_restore: WARNING:  database with OID 0 must be vacuumed within 
> 1000279 transactions
> HINT:  To avoid a database shutdown, execute a database-wide VACUUM in 
> that database.
> You might also need to commit or roll back old prepared transactions.
> pg_restore: WARNING:  database with OID 0 must be vacuumed within 
> 1000278 transactions
> HINT:  To avoid a database shutdown, execute a database-wide VACUUM in 
> that database.
> You might also need to commit or roll back old prepared transactions.
> 
> I've tried to do VACUUM FULL on my 9.6 cluster on all databases and then 
> retried the pg_upgrade - it failed in the same way.
> 
> Also to be noted, earlier this cluster was succesfully upgraded with 
> pg_upgrade using similar parameters from older versions (at least 2 
> times, something like 9.1 -> 9.3, 9.3 -> 9.6). The database is around 
> 700 GB and has very many pg_largeobjects in it.
> 
> What could be the reason of this and how can I perform my upgrade?
> 
> Thanks in advance,
> Alexander


-- 
Adrian Klaver
adrian.klaver@aklaver.com

Adrian Klaver <adrian.klaver@aklaver.com> writes:
> On 06/09/2018 03:46 AM, Alexander Shutyaev wrote:
>> The upgrade operation failed after several hours with the following error:
>> database is not accepting commands to avoid wraparound data loss in 
>> database with OID 0

> Do you know which database has an OID of 0?

Well, none do, so the correct question is what is passing an invalid
database OID to the code that's complaining.  This sure looks like a
bug, though I'm not sure we have enough info to locate it.

            regards, tom lane

I can provide you with more info. Please tell me what you need. I really hope
someone here can help me somehow solve or workaround this, because I really
need to migrate to v10 for its features.

Thanks in advance,
Alexander



--
Sent from: http://www.postgresql-archive.org/PostgreSQL-general-f1843780.html

On 06/10/2018 01:46 PM, Alexander Shutyaev wrote:
> I can provide you with more info. Please tell me what you need. I really hope
> someone here can help me somehow solve or workaround this, because I really
> need to migrate to v10 for its features.

Well as Tom Lane said a database will not have an OID of 0, so we need 
to know where that is coming from. On the oft chance it occurred in your 
9.6 cluster can you see if:

select oid, datname from pg_database;

in the 9.6 cluster shows an OID of 0.

Also:

1) OS and version you are using?

2) Where are you getting you Postgres from?:
a) Package
    Repo?
b) Source
    File source?

3) Where there any issues with 9.6 cluster before you tried to upgrade 
from it?


> 
> Thanks in advance,
> Alexander
> 
> 
> 
> --
> Sent from: http://www.postgresql-archive.org/PostgreSQL-general-f1843780.html
> 
> 


-- 
Adrian Klaver
adrian.klaver@aklaver.com

On 06/10/2018 02:09 PM, Alexander Shutyaev wrote:
> Some more notes on databses.
> 
> Although the pg_upgrade failed, I've decided to check the databases in 
> the new cluster (10.4). There is no database with oid 0 either. Also to 
> be noted that some system databases changed the oids while others 
> retained them.

If I am following the source code for pg_upgrade correctly that is 
expected. Pretty sure because the order of object creation is different.

> 
> And of my databases - sslentry. It had a very big oid (can that seem 
> strange?) and its oid has changed.

OID's are added at time of object creation so I would say the ssslentry 
database was created some time after the other databases in the 9.6 
cluster. Actually probably more accurate to say after 1016305714 - 
16400(bof db) objects that have OID's where created.

When the upgrade failed pg_upgrade should have pointed you at an error log.

Did it and is there anything useful there?

 From your OP post:

"You might also need to commit or roll back old prepared transactions."

Do you have any of those in the 9.6 cluster?

See:

https://www.postgresql.org/docs/10/static/view-pg-prepared-xacts.html

> 
> select oid, datname from pg_database;
>    oid  |  datname
> -------+-----------
>   13011 | template0
>   16400 | bof
>   13012 | postgres
>   16401 | sslentry
>       1 | template1
> (5 rows)
> 
> 



-- 
Adrian Klaver
adrian.klaver@aklaver.com

On 06/10/2018 02:45 PM, Alexander Shutyaev wrote:

Comments inline.

> The error log is like this. Here's its tail:

Is this the regular Postgres log or the pg_upgrade log which should be 
something like pg_upgrade_server.log?


> pg_restore: [archiver (db)] could not execute query: ERROR:  database is 
> not accepting commands to avoid wraparound data loss in database with OID 0
> HINT:  Stop the postmaster and vacuum that database in single-user mode.

How did you get into the 10 cluster to report on the database OID's and 
names?

> You might also need to commit or roll back old prepared transactions.
>      Command was: ALTER LARGE OBJECT 1740737402 OWNER TO bof_user;
> 
> Before that there is a lot of similar messages - the only things 
> chainging are the "executing BLOB nnn" number and "must be vacuumed 
> within nnn transactions" number.
> 

Which database has the large objects?

> As for the prepared transactions - no, I don't have them, our 
> application doesn't use this functionality.

Did you check this view to confirm?:

https://www.postgresql.org/docs/10/static/view-pg-prepared-xacts.html

Just trying to eliminate possibilities.



-- 
Adrian Klaver
adrian.klaver@aklaver.com

On 06/10/2018 11:46 PM, Alexander Shutyaev wrote:
>  >> Is this the regular Postgres log or the pg_upgrade log which should 
> be something like pg_upgrade_server.log?
> 
> This is the pg_upgrade_dump_16400.log.
> 
>>> How did you get into the 10 cluster to report on the database OID's and 
> names?
> 
> After the pg_upgrade failed I was able to start both clusters, so I 
> connected to the new 10.4 cluster and ran the query.

I am at a loss for an explanation. My thoughts:

1) The database with an  OID of 0 is a mystery, though it does not seem 
to be stopping the upgrade by itself.

2) The upgrade stops because of transaction ID wraparound, which is 
strange as that is not showing up in the 9.6 cluster I presume. You 
might want the queries found below on the 9.6 and 10 clusters to help 
figure this out:

https://www.postgresql.org/docs/10/static/routine-vacuuming.html#VACUUM-FOR-WRAPAROUND

SELECT c.oid::regclass as table_name,
        greatest(age(c.relfrozenxid),age(t.relfrozenxid)) as age
FROM pg_class c
LEFT JOIN pg_class t ON c.reltoastrelid = t.oid
WHERE c.relkind IN ('r', 'm');

SELECT datname, age(datfrozenxid) FROM pg_database;

> 
>>> Which database has the large objects?
> 
> bof (OID=16400). It is also effectively the only database that matters 
> here. The other one - sslentry only contains a couple of tables and a 
> dozen of records.
> 
>>> Did you check this view to confirm?
> 
> Yes, I did:
> 
> select * from pg_prepared_xacts;
>   transaction | gid | prepared | owner | database
> -------------+-----+----------+-------+----------
> (0 rows)
> 


-- 
Adrian Klaver
adrian.klaver@aklaver.com

Hi,

On 2018-06-09 13:46:16 +0300, Alexander Shutyaev wrote:
> Hello!
> 
> I've been trying to upgrade a postgresql cluster from 9.6 to 10. I've
> executed the pg_upgrade with the following options:
> 
>  /usr/lib/postgresql/10/bin/pg_upgrade -b /usr/lib/postgresql/9.6/bin/ -B
> /usr/lib/postgresql/10/bin/ -d /var/lib/postgresql/9.6/main -D
> /var/lib/postgresql/10/main -o ' -c
> config_file=/etc/postgresql/9.6/main/postgresql.conf' -O ' -c
> config_file=/etc/postgresql/10/main/postgresql.conf'
> 
> The upgrade operation failed after several hours with the following error:
> 
> database is not accepting commands to avoid wraparound data loss in
> database with OID 0
> 
> Earlier in the log there are a lot of messages like
> 
> pg_restore: executing BLOB 1740736966
> pg_restore: WARNING:  database with OID 0 must be vacuumed within 1000279
> transactions
> HINT:  To avoid a database shutdown, execute a database-wide VACUUM in that
> database.
> You might also need to commit or roll back old prepared transactions.

Since this happens in the version you're migrating too, I suggest
replacing the warning with a PANIC and then using the debugger to look
at the corefile generated (when using an appropriate ulimit).  It's not
immediately obvious why there'd not be correct knowledge about the
oldest database around. If you do so, please include 'p *ShmemVariableCache'
output.

Greetings,

Andres Freund

On 2018-06-09 15:52:26 -0400, Tom Lane wrote:
> Adrian Klaver <adrian.klaver@aklaver.com> writes:
> > On 06/09/2018 03:46 AM, Alexander Shutyaev wrote:
> >> The upgrade operation failed after several hours with the following error:
> >> database is not accepting commands to avoid wraparound data loss in 
> >> database with OID 0
> 
> > Do you know which database has an OID of 0?
> 
> Well, none do, so the correct question is what is passing an invalid
> database OID to the code that's complaining.  This sure looks like a
> bug, though I'm not sure we have enough info to locate it.

It sure looks like ShmemVariableCache->oldestXidDB isn't initialized.
As far as I remember we just initialize that from a checkpoint at
startup.   I suspect the issue is that pg_resetwal does:
    if (set_xid != 0)
    {
        ControlFile.checkPointCopy.nextXid = set_xid;

        /*
         * For the moment, just set oldestXid to a value that will force
         * immediate autovacuum-for-wraparound.  It's not clear whether adding
         * user control of this is useful, so let's just do something that's
         * reasonably safe.  The magic constant here corresponds to the
         * maximum allowed value of autovacuum_freeze_max_age.
         */
        ControlFile.checkPointCopy.oldestXid = set_xid - 2000000000;
        if (ControlFile.checkPointCopy.oldestXid < FirstNormalTransactionId)
            ControlFile.checkPointCopy.oldestXid += FirstNormalTransactionId;
        ControlFile.checkPointCopy.oldestXidDB = InvalidOid;
    }

but we have codepath that doesn't check for oldestXidDB being
InvalidOid.  Not great.

Greetings,

Andres Freund

Andres Freund <andres@anarazel.de> writes:
> I suspect the issue is that pg_resetwal does:
>     if (set_xid != 0)
>     {
>         ControlFile.checkPointCopy.nextXid = set_xid;

>         /*
>          * For the moment, just set oldestXid to a value that will force
>          * immediate autovacuum-for-wraparound.  It's not clear whether adding
>          * user control of this is useful, so let's just do something that's
>          * reasonably safe.  The magic constant here corresponds to the
>          * maximum allowed value of autovacuum_freeze_max_age.
>          */
>         ControlFile.checkPointCopy.oldestXid = set_xid - 2000000000;
>         if (ControlFile.checkPointCopy.oldestXid < FirstNormalTransactionId)
>             ControlFile.checkPointCopy.oldestXid += FirstNormalTransactionId;
>         ControlFile.checkPointCopy.oldestXidDB = InvalidOid;
>     }

> but we have codepath that doesn't check for oldestXidDB being
> InvalidOid.  Not great.

Hm, I think I'd define the problem as "pg_resetwal is violating the
expectation that oldestXidDB be valid".

However, this just explains the basically-cosmetic issue that the
complaint message mentions OID 0.  It doesn't really get us to the
answer to why Alexander is seeing a failure.  It might be useful
to see pg_controldata output for the old cluster, as well as
"select datname, datfrozenxid from pg_database" output from the
old cluster.

            regards, tom lane

On 2018-06-11 13:14:12 -0400, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
> > I suspect the issue is that pg_resetwal does:
> >     if (set_xid != 0)
> >     {
> >         ControlFile.checkPointCopy.nextXid = set_xid;
> 
> >         /*
> >          * For the moment, just set oldestXid to a value that will force
> >          * immediate autovacuum-for-wraparound.  It's not clear whether adding
> >          * user control of this is useful, so let's just do something that's
> >          * reasonably safe.  The magic constant here corresponds to the
> >          * maximum allowed value of autovacuum_freeze_max_age.
> >          */
> >         ControlFile.checkPointCopy.oldestXid = set_xid - 2000000000;
> >         if (ControlFile.checkPointCopy.oldestXid < FirstNormalTransactionId)
> >             ControlFile.checkPointCopy.oldestXid += FirstNormalTransactionId;
> >         ControlFile.checkPointCopy.oldestXidDB = InvalidOid;
> >     }
> 
> > but we have codepath that doesn't check for oldestXidDB being
> > InvalidOid.  Not great.
> 
> Hm, I think I'd define the problem as "pg_resetwal is violating the
> expectation that oldestXidDB be valid".

Well, what could it do otherwise?  ForceTransactionIdLimitUpdate()
currently does a syscache check for database existence. That'll just
return a lookup failure for InvalidOid, so we're reasonably good on that
front.

Using a hardcoded 2000000000 seems worse, will have funny results if
running with a smaller autovacuum_freeze_max_age...


> However, this just explains the basically-cosmetic issue that the
> complaint message mentions OID 0.  It doesn't really get us to the
> answer to why Alexander is seeing a failure.  It might be useful
> to see pg_controldata output for the old cluster, as well as
> "select datname, datfrozenxid from pg_database" output from the
> old cluster.

pg_upgrade starts the server with autovacuum disabled, I suspect
restoring all the large objects ends up using a lot of transaction
ids. GetNewTransactionId() should start autovacuum, but I'd guess that's
where things are going wrong for some reason.

Alexander, could you hack things up so autovacuum logging is enabled
(log_autovacuum_min_duration=0), and see whether it's triggered?

I'm not entirely clear why pg_restore appears to use a separate
transaction for each large object, surely exascerbating the problem.

Greetings,

Andres Freund

On 06/11/2018 11:32 AM, Alexander Shutyaev wrote:
> I'm back with more details.
> 
> First, I've deleted the smaller sslentry database, since I don't need 
> it, just so that it doesn't somehow spoil the picture. Now there is only 
> 1 user database - bof (OID=16400). After that I've ran the pg_upgrade on 
> a clean 10.4 cluster and it failed in the same way.
> 
> Now, the answers to your queries.
> 
>  >> 2) The upgrade stops because of transaction ID wraparound, which is 
> strange as that is not showing up in the 9.6 cluster I presume. You 
> might want the queries found below on the 9.6 and 10 clusters to help 
> figure this out:
> 
>  >> SELECT c.oid::regclass as table_name,
>  >>        greatest(age(c.relfrozenxid),age(t.relfrozenxid)) as age
>  >> FROM pg_class c
>  >> LEFT JOIN pg_class t ON c.reltoastrelid = t.oid
>  >> WHERE c.relkind IN ('r', 'm');
> 
> 9.6 result - see attached 96-query1.txt
> 10.4 result - see attached 104-query1.txt

I should have been clearer in my previous post, the above query is per 
database. From the query below I am going to say the above query was 
done on the bof database. Is that correct?

Given the below from 96-query2.txt:

template0 | 110588398

Can you run the table_name query in template0 in the 9.6 cluster?


> 
>  >> SELECT datname, age(datfrozenxid) FROM pg_database;
> 
> 9.6 result - see attached 96-query2.txt
> 10.4 result - see attached 104-query2.txt
> 
>  >>  It might be useful
>  >> to see pg_controldata output for the old cluster, as well as
>  >> "select datname, datfrozenxid from pg_database" output from the
>  >> old cluster.
> 
> for the query - see above, for pg_controldata:
> 
> 9.6 - see attached 96-pg_controldata.txt
> 10.4 - see attached 104-pg_controldata.txt
> 
>  >> Alexander, could you hack things up so autovacuum logging is enabled
>  >> (log_autovacuum_min_duration=0), and see whether it's triggered?
> 
> I'll be happy to, but that will require to run pg_upgrade once more and 
> that takes more that half a day and during this time clusters are not 
> available to me. Given the data I'm attaching it may happen that the 
> colleagues will want to see something else from my clusters or maybe 
> change some settings before running the pg_upgrade again. Therefore, 
> I'll wait 12 hours after this message in case there will be any more 
> requests and the I'll run the pg_upgrade again.
> 
> Thank you all for trying to solve this matter, this is much appreciated! :)
> 
> 2018-06-11 20:29 GMT+03:00 Andres Freund <andres@anarazel.de 
> <mailto:andres@anarazel.de>>:
> 
>     On 2018-06-11 13:14:12 -0400, Tom Lane wrote:
>     > Andres Freund <andres@anarazel.de <mailto:andres@anarazel.de>> writes:
>     > > I suspect the issue is that pg_resetwal does:
>     > >     if (set_xid != 0)
>     > >     {
>     > >             ControlFile.checkPointCopy.nextXid = set_xid;
>     > 
>     > >             /*
>     > >              * For the moment, just set oldestXid to a value that will force
>     > >              * immediate autovacuum-for-wraparound.  It's not clear whether adding
>     > >              * user control of this is useful, so let's just do something that's
>     > >              * reasonably safe.  The magic constant here corresponds to the
>     > >              * maximum allowed value of autovacuum_freeze_max_age.
>     > >              */
>     > >             ControlFile.checkPointCopy.oldestXid = set_xid - 2000000000;
>     > >             if (ControlFile.checkPointCopy.oldestXid < FirstNormalTransactionId)
>     > >                     ControlFile.checkPointCopy.oldestXid += FirstNormalTransactionId;
>     > >             ControlFile.checkPointCopy.oldestXidDB = InvalidOid;
>     > >     }
>     > 
>     > > but we have codepath that doesn't check for oldestXidDB being
>     > > InvalidOid.  Not great.
>     > 
>     > Hm, I think I'd define the problem as "pg_resetwal is violating the
>     > expectation that oldestXidDB be valid".
> 
>     Well, what could it do otherwise?  ForceTransactionIdLimitUpdate()
>     currently does a syscache check for database existence. That'll just
>     return a lookup failure for InvalidOid, so we're reasonably good on that
>     front.
> 
>     Using a hardcoded 2000000000 seems worse, will have funny results if
>     running with a smaller autovacuum_freeze_max_age...
> 
> 
>     > However, this just explains the basically-cosmetic issue that the
>     > complaint message mentions OID 0.  It doesn't really get us to the
>     > answer to why Alexander is seeing a failure.  It might be useful
>     > to see pg_controldata output for the old cluster, as well as
>     > "select datname, datfrozenxid from pg_database" output from the
>     > old cluster.
> 
>     pg_upgrade starts the server with autovacuum disabled, I suspect
>     restoring all the large objects ends up using a lot of transaction
>     ids. GetNewTransactionId() should start autovacuum, but I'd guess that's
>     where things are going wrong for some reason.
> 
>     Alexander, could you hack things up so autovacuum logging is enabled
>     (log_autovacuum_min_duration=0), and see whether it's triggered?
> 
>     I'm not entirely clear why pg_restore appears to use a separate
>     transaction for each large object, surely exascerbating the problem.
> 
>     Greetings,
> 
>     Andres Freund
> 
> 


-- 
Adrian Klaver
adrian.klaver@aklaver.com

    Andres Freund wrote:

> I'm not entirely clear why pg_restore appears to use a separate
> transaction for each large object, surely exascerbating the problem.

To make sure that per-object locks don't fill up the shared
lock table?
There might be hundreds of thousands of large objects.
If it had to restore N objects per transaction, would it know
how to compute N that is large enough to be effective
and small enough not to exhaust the shared table?

Best regards,
--
Daniel Vérité
PostgreSQL-powered mailer: http://www.manitou-mail.org
Twitter: @DanielVerite

On Tue, Jun 26, 2018 at 8:38 AM Alexander Shutyaev <shutyaev@gmail.com> wrote:
>
> Hello again,
>
> I've performed another test - I've migrated to the new cluster using dump restore: pg_dumpall | psql. It went well,
althoughit took 6 days while pg_upgrade usually took a night.
 
>
> Is there any hope the issue with pg_upgrade can be resolved? If not, could you give me some hints as to how can I
decreasetime needed for pg_dumpall | psql?
 

If you use pg_dump with the custom format you can use pg_restore
--jobs=n to do a parallel restore.

I think you can do pg_dumpall --globals-only to restore users, and
then do a pg_dump per database.

Groeten, Arjen

    Alexander Shutyaev wrote:

> Is there any hope the issue with pg_upgrade can be resolved? If not,
> could you give me some hints as to how can I decrease time needed
> for pg_dumpall | psql?

Not sure about the larger problem, but for the part about having more
than 1 million large objects that get committed individually, setting
fsync=off for the reload phase is likely to help a lot.

Just don't forget to turn it on again when it's done. See
https://blog.2ndquadrant.com/postgresql-fsync-off-warning-in-config-file/


Best regards,
--
Daniel Vérité
PostgreSQL-powered mailer: http://www.manitou-mail.org
Twitter: @DanielVerite

On Sat, Mar 13, 2021 at 08:43:54AM -0500, Jan Wieck wrote:
> On 3/12/21 8:30 PM, Michael Paquier wrote:
> > Hi Jan,
> > 
> > On Fri, Mar 12, 2021 at 06:13:33PM -0500, Jan Wieck wrote:
> > > One of the things in my way is that when using pg_resetwal to put the
> > > NextXID way into the future (to push the old cluster close to wraparound for
> > > example), the postmaster won't start because it doesn't have the pg_xact
> > > files for that around. Should pg_resetwal create the files in the gap
> > > between the old NextXID and the new one?
> > 
> > I think that you should add this patch to the next commit fest to
> > track it properly:
> > https://commitfest.postgresql.org/33/
> > --
> > Michael
> > 
> 
> Actually this is the wrong patch (this one is for PG-12, not for HEAD). Will
> update later today.
> 
> But yes, putting it into the next commitfest after initial discussion is the
> plan.

Uh, were either of these things done?

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EDB                                      https://enterprisedb.com

  If only the physical world exists, free will is an illusion.

On 4/30/21 3:32 PM, Bruce Momjian wrote:
> On Sat, Mar 13, 2021 at 08:43:54AM -0500, Jan Wieck wrote:
>> On 3/12/21 8:30 PM, Michael Paquier wrote:
>> > Hi Jan,
>> > 
>> > On Fri, Mar 12, 2021 at 06:13:33PM -0500, Jan Wieck wrote:
>> > > One of the things in my way is that when using pg_resetwal to put the
>> > > NextXID way into the future (to push the old cluster close to wraparound for
>> > > example), the postmaster won't start because it doesn't have the pg_xact
>> > > files for that around. Should pg_resetwal create the files in the gap
>> > > between the old NextXID and the new one?
>> > 
>> > I think that you should add this patch to the next commit fest to
>> > track it properly:
>> > https://commitfest.postgresql.org/33/
>> > --
>> > Michael
>> > 
>> 
>> Actually this is the wrong patch (this one is for PG-12, not for HEAD). Will
>> update later today.
>> 
>> But yes, putting it into the next commitfest after initial discussion is the
>> plan.
> 
> Uh, were either of these things done?
> 

Not yet, but I will enter it so that we can get it into 15 for sure.


Regards, Jan

-- 
Jan Wieck
Postgres User since 1994

On Mon, May 03, 2021 at 11:10:44AM -0400, Jan Wieck wrote:
> Not yet, but I will enter it so that we can get it into 15 for sure.

I may be missing something but this is not listed:
https://commitfest.postgresql.org/33/

Could you add it to the CF app please?  There are so many patches and
discussions that this would easily get lost if you don't register it.
And from what I can see having a discussion on this matter looks
adapted to me.
--
Michael