logical decoding bug when mapped relation with toast contents is rewritten repeatedly

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

logical decoding bug when mapped relation with toast contents is rewritten repeatedly

Andres Freund
Hi,

(Tomas, CCing you because you IIRC mentioned encountered an issue like
this)

I just spent quite a while debugging an issue where running logical
decoding yielded a:
ERROR:  could not map filenode "base/$X/$Y" to relation OID
error.

After discarding like 30 different theories, I have found the cause:

During rewrites (i.e. VACUUM FULL / CLUSTER) of a mapped relation with a
toast table with actual live toasted tuples (pg_proc in my case and
henceforth) heap inserts with the toast data happen into the new toast
relation, triggered by:

static void
raw_heap_insert(RewriteState state, HeapTuple tup)
...
        /*
         * If the new tuple is too big for storage or contains already toasted
         * out-of-line attributes from some other relation, invoke the toaster.
         *
         * Note: below this point, heaptup is the data we actually intend to store
         * into the relation; tup is the caller's original untoasted data.
         */
        if (state->rs_new_rel->rd_rel->relkind == RELKIND_TOASTVALUE)
        {
                /* toast table entries should never be recursively toasted */
                Assert(!HeapTupleHasExternal(tup));
                heaptup = tup;
        }
        else if (HeapTupleHasExternal(tup) || tup->t_len > TOAST_TUPLE_THRESHOLD)
                heaptup = toast_insert_or_update(state->rs_new_rel, tup, NULL,
                                                                                 HEAP_INSERT_SKIP_FSM |
                                                                                 (state->rs_use_wal ?
                                                                                  0 : HEAP_INSERT_SKIP_WAL));
        else
                heaptup = tup;


At that point the new toast relation does *NOT* appear to be a system
catalog, it's just appears as an "independent" table.  Therefore we do
not trigger, in heap_insert():

/*
 * RelationIsLogicallyLogged
 * True if we need to log enough information to extract the data from the
 * WAL stream.
 *
 * We don't log information for unlogged tables (since they don't WAL log
 * anyway) and for system tables (their content is hard to make sense of, and
 * it would complicate decoding slightly for little gain). Note that we *do*
 * log information for user defined catalog tables since they presumably are
 * interesting to the user...
 */
#define RelationIsLogicallyLogged(relation) \
        (XLogLogicalInfoActive() && \
         RelationNeedsWAL(relation) && \
         !IsCatalogRelation(relation))

                /*
                 * For logical decoding, we need the tuple even if we're doing a full
                 * page write, so make sure it's included even if we take a full-page
                 * image. (XXX We could alternatively store a pointer into the FPW).
                 */
                if (RelationIsLogicallyLogged(relation))
                {
                        xlrec.flags |= XLH_INSERT_CONTAINS_NEW_TUPLE;
                        bufflags |= REGBUF_KEEP_DATA;
                }

i.e. the inserted toast tuple will be marked as
XLH_INSERT_CONTAINS_NEW_TUPLE - which it shouldn't, because it's a
system table. Which we currently do not allow do be logically decoded.

That normally ends up being harmless, because ReorderBufferCommit() has the
following check:
                                        if (!RelationIsLogicallyLogged(relation))
                                                goto change_done;

but to reach that check, we first have to map the relfilenode from the
WAL to the corresponding OID:
                                        reloid = RelidByRelfilenode(change->data.tp.relnode.spcNode,
                                                                                                change->data.tp.relnode.relNode);

That works correctly if there's only one rewrite - the relmapper
contains the data for the new toast table.  But if there's been *two*
consecutive rewrites, the relmapper *does not* contain the intermediary
relfilenode of pg_proc.  There's no such problem for non-mapped tables,
because historic snapshots allow us to access the relevant data, but the
relmapper isn't mvcc.

Therefore the catalog-rewrite escape hatch of:
                                        /*
                                         * Catalog tuple without data, emitted while catalog was
                                         * in the process of being rewritten.
                                         */
                                        if (reloid == InvalidOid &&
                                                change->data.tp.newtuple == NULL &&
                                                change->data.tp.oldtuple == NULL)
                                                goto change_done;
does not trigger and we run into:
                                        else if (reloid == InvalidOid)
                                                elog(ERROR, "could not map filenode \"%s\" to relation OID",
                                                         relpathperm(change->data.tp.relnode,
                                                                                 MAIN_FORKNUM));


commenting out this error / converting it into a warning makes this case
harmless, but could obviously be problematic in other scenarios.


I suspect the proper fix would be to have a new HEAP_INSERT_NO_LOGICAL
option, and specify that in raw_heap_insert() iff
RelationIsLogicallyLogged(state->rs_old_rel) or something like that.

Attached is a *prototype* patch of that approach.  Without the code
level changes the addition to test_decoding's rewrite.sql trigger the
bug, after it they're fixed.


The only reason the scenario I was debugging hit this was that there was
a cluster wide VACUUM FULL a couple times a day, and replication was
several hours behind due to slow network / receiving side.


Now I'm having a beer outside.


Greetings,

Andres Freund

decoding-fix.diff (23K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: logical decoding bug when mapped relation with toast contents is rewritten repeatedly

Tomas Vondra-4
Hi,

On 09/14/2018 04:10 AM, Andres Freund wrote:
> Hi,
>
> (Tomas, CCing you because you IIRC mentioned encountered an issue like
> this)
>

I might have mentioned an issue with this symptom recently, but that
turned out to be already fixed by da10d6a8a9 (before the minor version
with that fix got released).

> I just spent quite a while debugging an issue where running logical
> decoding yielded a:
> ERROR:  could not map filenode "base/$X/$Y" to relation OID
> error.
>
> After discarding like 30 different theories, I have found the cause:
>

Yeah. These issues are not exactly trivial to investigate ;-)

> During rewrites (i.e. VACUUM FULL / CLUSTER) of a mapped relation with a
> toast table with actual live toasted tuples (pg_proc in my case and
> henceforth) heap inserts with the toast data happen into the new toast
> relation, triggered by:
>
> static void
> raw_heap_insert(RewriteState state, HeapTuple tup)
> ...
> /*
> * If the new tuple is too big for storage or contains already toasted
> * out-of-line attributes from some other relation, invoke the toaster.
> *
> * Note: below this point, heaptup is the data we actually intend to store
> * into the relation; tup is the caller's original untoasted data.
> */
> if (state->rs_new_rel->rd_rel->relkind == RELKIND_TOASTVALUE)
> {
> /* toast table entries should never be recursively toasted */
> Assert(!HeapTupleHasExternal(tup));
> heaptup = tup;
> }
> else if (HeapTupleHasExternal(tup) || tup->t_len > TOAST_TUPLE_THRESHOLD)
> heaptup = toast_insert_or_update(state->rs_new_rel, tup, NULL,
> HEAP_INSERT_SKIP_FSM |
> (state->rs_use_wal ?
>  0 : HEAP_INSERT_SKIP_WAL));
> else
> heaptup = tup;
>
>
> At that point the new toast relation does *NOT* appear to be a system
> catalog, it's just appears as an "independent" table.  Therefore we do
> not trigger, in heap_insert():
>

Hmm, can't we change that? Recognizing the new TOAST table as a catalog
would fix the issue, no?

> /*
>  * RelationIsLogicallyLogged
>  * True if we need to log enough information to extract the data from the
>  * WAL stream.
>  *
>  * We don't log information for unlogged tables (since they don't WAL log
>  * anyway) and for system tables (their content is hard to make sense of, and
>  * it would complicate decoding slightly for little gain). Note that we *do*
>  * log information for user defined catalog tables since they presumably are
>  * interesting to the user...
>  */
> #define RelationIsLogicallyLogged(relation) \
> (XLogLogicalInfoActive() && \
> RelationNeedsWAL(relation) && \
> !IsCatalogRelation(relation))
>
> /*
> * For logical decoding, we need the tuple even if we're doing a full
> * page write, so make sure it's included even if we take a full-page
> * image. (XXX We could alternatively store a pointer into the FPW).
> */
> if (RelationIsLogicallyLogged(relation))
> {
> xlrec.flags |= XLH_INSERT_CONTAINS_NEW_TUPLE;
> bufflags |= REGBUF_KEEP_DATA;
> }
>
> i.e. the inserted toast tuple will be marked as
> XLH_INSERT_CONTAINS_NEW_TUPLE - which it shouldn't, because it's a
> system table. Which we currently do not allow do be logically decoded.
>
> That normally ends up being harmless, because ReorderBufferCommit() has the
> following check:
> if (!RelationIsLogicallyLogged(relation))
> goto change_done;
>
> but to reach that check, we first have to map the relfilenode from the
> WAL to the corresponding OID:
> reloid = RelidByRelfilenode(change->data.tp.relnode.spcNode,
> change->data.tp.relnode.relNode);
>
> That works correctly if there's only one rewrite - the relmapper
> contains the data for the new toast table.  But if there's been *two*
> consecutive rewrites, the relmapper *does not* contain the intermediary
> relfilenode of pg_proc.  There's no such problem for non-mapped tables,
> because historic snapshots allow us to access the relevant data, but the
> relmapper isn't mvcc.
>
> Therefore the catalog-rewrite escape hatch of:
> /*
> * Catalog tuple without data, emitted while catalog was
> * in the process of being rewritten.
> */
> if (reloid == InvalidOid &&
> change->data.tp.newtuple == NULL &&
> change->data.tp.oldtuple == NULL)
> goto change_done;
> does not trigger and we run into:
> else if (reloid == InvalidOid)
> elog(ERROR, "could not map filenode \"%s\" to relation OID",
> relpathperm(change->data.tp.relnode,
> MAIN_FORKNUM));
>
>
> commenting out this error / converting it into a warning makes this case
> harmless, but could obviously be problematic in other scenarios.
>

Yeah, that seems like a bad idea. That error already caught a couple of
bugs (including da10d6a8a9 and this one), and I have a hunch those are
not the last ones.

>
> I suspect the proper fix would be to have a new HEAP_INSERT_NO_LOGICAL
> option, and specify that in raw_heap_insert() iff
> RelationIsLogicallyLogged(state->rs_old_rel) or something like that.
>
> Attached is a *prototype* patch of that approach.  Without the code
> level changes the addition to test_decoding's rewrite.sql trigger the
> bug, after it they're fixed.
>
>
> The only reason the scenario I was debugging hit this was that there was
> a cluster wide VACUUM FULL a couple times a day, and replication was
> several hours behind due to slow network / receiving side.
>
>
> Now I'm having a beer outside.
>

After discarding 30 theories? Have two.


regards

--
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Reply | Threaded
Open this post in threaded view
|

Re: logical decoding bug when mapped relation with toast contents is rewritten repeatedly

Andres Freund
Hi,

On 2018-09-14 16:13:46 +0200, Tomas Vondra wrote:

> >
> > I suspect the proper fix would be to have a new HEAP_INSERT_NO_LOGICAL
> > option, and specify that in raw_heap_insert() iff
> > RelationIsLogicallyLogged(state->rs_old_rel) or something like that.
> >
> > Attached is a *prototype* patch of that approach.  Without the code
> > level changes the addition to test_decoding's rewrite.sql trigger the
> > bug, after it they're fixed.
> >
> >
> > The only reason the scenario I was debugging hit this was that there was
> > a cluster wide VACUUM FULL a couple times a day, and replication was
> > several hours behind due to slow network / receiving side.
> >
> >
> > Now I'm having a beer outside.

> Yeah, that seems like a bad idea. That error already caught a couple of
> bugs (including da10d6a8a9 and this one), and I have a hunch those are
> not the last ones.

One problem with this is that that means upgrading won't fix an existing
instance of the problem, but turning the ERROR into a WARNING would. I
personally think that's *NOT* enough justification for relaxing the
error, given that recreating the slot would fix the issue, but I see how
other people can reasonably differ.  I can't really see a reasonably
complex approach that solves the issue in a "cake but have it too"
way...


> After discarding 30 theories? Have two.

I will neither confirm nor deny. ;)


Greetings,

Andres Freund