tablesync.c File Reference

#include "postgres.h"
#include "access/table.h"
#include "access/xact.h"
#include "catalog/indexing.h"
#include "catalog/pg_subscription_rel.h"
#include "catalog/pg_type.h"
#include "commands/copy.h"
#include "miscadmin.h"
#include "nodes/makefuncs.h"
#include "parser/parse_relation.h"
#include "pgstat.h"
#include "replication/logicallauncher.h"
#include "replication/logicalrelation.h"
#include "replication/logicalworker.h"
#include "replication/origin.h"
#include "replication/slot.h"
#include "replication/walreceiver.h"
#include "replication/worker_internal.h"
#include "storage/ipc.h"
#include "storage/lmgr.h"
#include "utils/acl.h"
#include "utils/array.h"
#include "utils/builtins.h"
#include "utils/lsyscache.h"
#include "utils/rls.h"
#include "utils/snapmgr.h"
#include "utils/syscache.h"
#include "utils/usercontext.h"

Include dependency graph for tablesync.c:

Go to the source code of this file.

Functions
static bool	wait_for_table_state_change (Oid relid, char expected_state)
static bool	wait_for_worker_state_change (char expected_state)
void	ProcessSyncingTablesForSync (XLogRecPtr current_lsn)
void	ProcessSyncingTablesForApply (XLogRecPtr current_lsn)
static List *	make_copy_attnamelist (LogicalRepRelMapEntry *rel)
static int	copy_read_data (void *outbuf, int minread, int maxread)
static void	fetch_remote_table_info (char *nspname, char *relname, LogicalRepRelation *lrel, List **qual, bool *gencol_published)
static void	copy_table (Relation rel)
void	ReplicationSlotNameForTablesync (Oid suboid, Oid relid, char *syncslotname, Size szslot)
static char *	LogicalRepSyncTableStart (XLogRecPtr *origin_startpos)
static void	start_table_sync (XLogRecPtr *origin_startpos, char **slotname)
static void	run_tablesync_worker (void)
void	TableSyncWorkerMain (Datum main_arg)
bool	AllTablesyncsReady (void)
bool	HasSubscriptionTablesCached (void)
void	UpdateTwoPhaseState (Oid suboid, char new_state)

Variables
List *	table_states_not_ready = NIL
static StringInfo	copybuf = NULL

Function Documentation

AllTablesyncsReady()

bool AllTablesyncsReady

(

void

)

Definition at line 1598 of file tablesync.c.

{
    bool        started_tx;
    bool        has_tables;
 
    /* We need up-to-date sync state info for subscription tables here. */
    FetchRelationStates(&has_tables, NULL, &started_tx);
 
    if (started_tx)
    {
        CommitTransactionCommand();
        pgstat_report_stat(true);
    }
 
    /*
     * Return false when there are no tables in subscription or not all tables
     * are in ready state; true otherwise.
     */
    return has_tables && (table_states_not_ready == NIL);
}

References CommitTransactionCommand(), fb(), FetchRelationStates(), NIL, pgstat_report_stat(), and table_states_not_ready.

Referenced by pa_can_start(), ProcessSyncingTablesForApply(), run_apply_worker(), and wait_for_local_flush().

copy_read_data()

static int copy_read_data ( void *  outbuf, int  minread, int  maxread  )

static

Definition at line 646 of file tablesync.c.

{
    int         bytesread = 0;
    int         avail;
 
    /* If there are some leftover data from previous read, use it. */
    avail = copybuf->len - copybuf->cursor;
    if (avail)
    {
        if (avail > maxread)
            avail = maxread;
        memcpy(outbuf, &copybuf->data[copybuf->cursor], avail);
        copybuf->cursor += avail;
        maxread -= avail;
        bytesread += avail;
    }
 
    while (maxread > 0 && bytesread < minread)
    {
        pgsocket    fd = PGINVALID_SOCKET;
        int         len;
        char       *buf = NULL;
 
        for (;;)
        {
            /* Try read the data. */
            len = walrcv_receive(LogRepWorkerWalRcvConn, &buf, &fd);
 
            CHECK_FOR_INTERRUPTS();
 
            if (len == 0)
                break;
            else if (len < 0)
                return bytesread;
            else
            {
                /* Process the data */
                copybuf->data = buf;
                copybuf->len = len;
                copybuf->cursor = 0;
 
                avail = copybuf->len - copybuf->cursor;
                if (avail > maxread)
                    avail = maxread;
                memcpy(outbuf, &copybuf->data[copybuf->cursor], avail);
                outbuf = (char *) outbuf + avail;
                copybuf->cursor += avail;
                maxread -= avail;
                bytesread += avail;
            }
 
            if (maxread <= 0 || bytesread >= minread)
                return bytesread;
        }
 
        /*
         * Wait for more data or latch.
         */
        (void) WaitLatchOrSocket(MyLatch,
                                 WL_SOCKET_READABLE | WL_LATCH_SET |
                                 WL_TIMEOUT | WL_EXIT_ON_PM_DEATH,
                                 fd, 1000L, WAIT_EVENT_LOGICAL_SYNC_DATA);
 
        ResetLatch(MyLatch);
    }
 
    return bytesread;
}

References buf, CHECK_FOR_INTERRUPTS, copybuf, StringInfoData::cursor, StringInfoData::data, fb(), fd(), StringInfoData::len, len, LogRepWorkerWalRcvConn, MyLatch, PGINVALID_SOCKET, ResetLatch(), WaitLatchOrSocket(), walrcv_receive, WL_EXIT_ON_PM_DEATH, WL_LATCH_SET, WL_SOCKET_READABLE, and WL_TIMEOUT.

Referenced by copy_table().

copy_table()

static void copy_table ( Relation  rel )

static

Definition at line 1043 of file tablesync.c.

{
    LogicalRepRelMapEntry *relmapentry;
    LogicalRepRelation lrel;
    List       *qual = NIL;
    WalRcvExecResult *res;
    StringInfoData cmd;
    CopyFromState cstate;
    List       *attnamelist;
    ParseState *pstate;
    List       *options = NIL;
    bool        gencol_published = false;
 
    /* Get the publisher relation info. */
    fetch_remote_table_info(get_namespace_name(RelationGetNamespace(rel)),
                            RelationGetRelationName(rel), &lrel, &qual,
                            &gencol_published);
 
    /* Put the relation into relmap. */
    logicalrep_relmap_update(&lrel);
 
    /* Map the publisher relation to local one. */
    relmapentry = logicalrep_rel_open(lrel.remoteid, NoLock);
    Assert(rel == relmapentry->localrel);
 
    /* Start copy on the publisher. */
    initStringInfo(&cmd);
 
    /* Regular or partitioned table with no row filter or generated columns */
    if ((lrel.relkind == RELKIND_RELATION || lrel.relkind == RELKIND_PARTITIONED_TABLE)
        && qual == NIL && !gencol_published)
    {
        appendStringInfo(&cmd, "COPY %s",
                         quote_qualified_identifier(lrel.nspname, lrel.relname));
 
        /* If the table has columns, then specify the columns */
        if (lrel.natts)
        {
            appendStringInfoString(&cmd, " (");
 
            /*
             * XXX Do we need to list the columns in all cases? Maybe we're
             * replicating all columns?
             */
            for (int i = 0; i < lrel.natts; i++)
            {
                if (i > 0)
                    appendStringInfoString(&cmd, ", ");
 
                appendStringInfoString(&cmd, quote_identifier(lrel.attnames[i]));
            }
 
            appendStringInfoChar(&cmd, ')');
        }
 
        appendStringInfoString(&cmd, " TO STDOUT");
    }
    else
    {
        /*
         * For non-tables and tables with row filters, we need to do COPY
         * (SELECT ...), but we can't just do SELECT * because we may need to
         * copy only subset of columns including generated columns. For tables
         * with any row filters, build a SELECT query with OR'ed row filters
         * for COPY.
         *
         * We also need to use this same COPY (SELECT ...) syntax when
         * generated columns are published, because copy of generated columns
         * is not supported by the normal COPY.
         */
        appendStringInfoString(&cmd, "COPY (SELECT ");
        for (int i = 0; i < lrel.natts; i++)
        {
            appendStringInfoString(&cmd, quote_identifier(lrel.attnames[i]));
            if (i < lrel.natts - 1)
                appendStringInfoString(&cmd, ", ");
        }
 
        appendStringInfoString(&cmd, " FROM ");
 
        /*
         * For regular tables, make sure we don't copy data from a child that
         * inherits the named table as those will be copied separately.
         */
        if (lrel.relkind == RELKIND_RELATION)
            appendStringInfoString(&cmd, "ONLY ");
 
        appendStringInfoString(&cmd, quote_qualified_identifier(lrel.nspname, lrel.relname));
        /* list of OR'ed filters */
        if (qual != NIL)
        {
            ListCell   *lc;
            char       *q = strVal(linitial(qual));
 
            appendStringInfo(&cmd, " WHERE %s", q);
            for_each_from(lc, qual, 1)
            {
                q = strVal(lfirst(lc));
                appendStringInfo(&cmd, " OR %s", q);
            }
            list_free_deep(qual);
        }
 
        appendStringInfoString(&cmd, ") TO STDOUT");
    }
 
    /*
     * Prior to v16, initial table synchronization will use text format even
     * if the binary option is enabled for a subscription.
     */
    if (walrcv_server_version(LogRepWorkerWalRcvConn) >= 160000 &&
        MySubscription->binary)
    {
        appendStringInfoString(&cmd, " WITH (FORMAT binary)");
        options = list_make1(makeDefElem("format",
                                         (Node *) makeString("binary"), -1));
    }
 
    res = walrcv_exec(LogRepWorkerWalRcvConn, cmd.data, 0, NULL);
    pfree(cmd.data);
    if (res->status != WALRCV_OK_COPY_OUT)
        ereport(ERROR,
                (errcode(ERRCODE_CONNECTION_FAILURE),
                 errmsg("could not start initial contents copy for table \"%s.%s\": %s",
                        lrel.nspname, lrel.relname, res->err)));
    walrcv_clear_result(res);
 
    copybuf = makeStringInfo();
 
    pstate = make_parsestate(NULL);
    (void) addRangeTableEntryForRelation(pstate, rel, AccessShareLock,
                                         NULL, false, false);
 
    attnamelist = make_copy_attnamelist(relmapentry);
    cstate = BeginCopyFrom(pstate, rel, NULL, NULL, false, copy_read_data, attnamelist, options);
 
    /* Do the copy */
    (void) CopyFrom(cstate);
 
    logicalrep_rel_close(relmapentry, NoLock);
}

References AccessShareLock, addRangeTableEntryForRelation(), appendStringInfo(), appendStringInfoChar(), appendStringInfoString(), Assert, BeginCopyFrom(), Subscription::binary, copy_read_data(), copybuf, CopyFrom(), StringInfoData::data, ereport, WalRcvExecResult::err, errcode(), errmsg(), ERROR, fb(), fetch_remote_table_info(), for_each_from, get_namespace_name(), i, initStringInfo(), lfirst, linitial, list_free_deep(), list_make1, LogicalRepRelMapEntry::localrel, logicalrep_rel_close(), logicalrep_rel_open(), logicalrep_relmap_update(), LogRepWorkerWalRcvConn, make_copy_attnamelist(), make_parsestate(), makeDefElem(), makeString(), makeStringInfo(), MySubscription, NIL, NoLock, pfree(), quote_identifier(), quote_qualified_identifier(), RelationGetNamespace, RelationGetRelationName, WalRcvExecResult::status, strVal, walrcv_clear_result(), walrcv_exec, WALRCV_OK_COPY_OUT, and walrcv_server_version.

Referenced by LogicalRepSyncTableStart().

fetch_remote_table_info()

static void fetch_remote_table_info ( char *  nspname, char *  relname, LogicalRepRelation *  lrel, List **  qual, bool *  gencol_published  )

static

Definition at line 725 of file tablesync.c.

{
    WalRcvExecResult *res;
    StringInfoData cmd;
    TupleTableSlot *slot;
    Oid         tableRow[] = {OIDOID, CHAROID, CHAROID};
    Oid         attrRow[] = {INT2OID, TEXTOID, OIDOID, BOOLOID, BOOLOID};
    Oid         qualRow[] = {TEXTOID};
    bool        isnull;
    int         natt;
    StringInfo  pub_names = NULL;
    Bitmapset  *included_cols = NULL;
    int         server_version = walrcv_server_version(LogRepWorkerWalRcvConn);
 
    lrel->nspname = nspname;
    lrel->relname = relname;
 
    /* First fetch Oid and replica identity. */
    initStringInfo(&cmd);
    appendStringInfo(&cmd, "SELECT c.oid, c.relreplident, c.relkind"
                     "  FROM pg_catalog.pg_class c"
                     "  INNER JOIN pg_catalog.pg_namespace n"
                     "        ON (c.relnamespace = n.oid)"
                     " WHERE n.nspname = %s"
                     "   AND c.relname = %s",
                     quote_literal_cstr(nspname),
                     quote_literal_cstr(relname));
    res = walrcv_exec(LogRepWorkerWalRcvConn, cmd.data,
                      lengthof(tableRow), tableRow);
 
    if (res->status != WALRCV_OK_TUPLES)
        ereport(ERROR,
                (errcode(ERRCODE_CONNECTION_FAILURE),
                 errmsg("could not fetch table info for table \"%s.%s\" from publisher: %s",
                        nspname, relname, res->err)));
 
    slot = MakeSingleTupleTableSlot(res->tupledesc, &TTSOpsMinimalTuple);
    if (!tuplestore_gettupleslot(res->tuplestore, true, false, slot))
        ereport(ERROR,
                (errcode(ERRCODE_UNDEFINED_OBJECT),
                 errmsg("table \"%s.%s\" not found on publisher",
                        nspname, relname)));
 
    lrel->remoteid = DatumGetObjectId(slot_getattr(slot, 1, &isnull));
    Assert(!isnull);
    lrel->replident = DatumGetChar(slot_getattr(slot, 2, &isnull));
    Assert(!isnull);
    lrel->relkind = DatumGetChar(slot_getattr(slot, 3, &isnull));
    Assert(!isnull);
 
    ExecDropSingleTupleTableSlot(slot);
    walrcv_clear_result(res);
 
 
    /*
     * Get column lists for each relation.
     *
     * We need to do this before fetching info about column names and types,
     * so that we can skip columns that should not be replicated.
     */
    if (server_version >= 150000)
    {
        WalRcvExecResult *pubres;
        TupleTableSlot *tslot;
        Oid         attrsRow[] = {INT2VECTOROID};
 
        /* Build the pub_names comma-separated string. */
        pub_names = makeStringInfo();
        GetPublicationsStr(MySubscription->publications, pub_names, true);
 
        /*
         * Fetch info about column lists for the relation (from all the
         * publications).
         */
        resetStringInfo(&cmd);
        appendStringInfo(&cmd,
                         "SELECT DISTINCT"
                         "  (CASE WHEN (array_length(gpt.attrs, 1) = c.relnatts)"
                         "   THEN NULL ELSE gpt.attrs END)"
                         "  FROM pg_publication p,"
                         "  LATERAL pg_get_publication_tables(p.pubname) gpt,"
                         "  pg_class c"
                         " WHERE gpt.relid = %u AND c.oid = gpt.relid"
                         "   AND p.pubname IN ( %s )",
                         lrel->remoteid,
                         pub_names->data);
 
        pubres = walrcv_exec(LogRepWorkerWalRcvConn, cmd.data,
                             lengthof(attrsRow), attrsRow);
 
        if (pubres->status != WALRCV_OK_TUPLES)
            ereport(ERROR,
                    (errcode(ERRCODE_CONNECTION_FAILURE),
                     errmsg("could not fetch column list info for table \"%s.%s\" from publisher: %s",
                            nspname, relname, pubres->err)));
 
        /*
         * We don't support the case where the column list is different for
         * the same table when combining publications. See comments atop
         * fetch_relation_list. So there should be only one row returned.
         * Although we already checked this when creating the subscription, we
         * still need to check here in case the column list was changed after
         * creating the subscription and before the sync worker is started.
         */
        if (tuplestore_tuple_count(pubres->tuplestore) > 1)
            ereport(ERROR,
                    errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
                    errmsg("cannot use different column lists for table \"%s.%s\" in different publications",
                           nspname, relname));
 
        /*
         * Get the column list and build a single bitmap with the attnums.
         *
         * If we find a NULL value, it means all the columns should be
         * replicated.
         */
        tslot = MakeSingleTupleTableSlot(pubres->tupledesc, &TTSOpsMinimalTuple);
        if (tuplestore_gettupleslot(pubres->tuplestore, true, false, tslot))
        {
            Datum       cfval = slot_getattr(tslot, 1, &isnull);
 
            if (!isnull)
            {
                ArrayType  *arr;
                int         nelems;
                int16      *elems;
 
                arr = DatumGetArrayTypeP(cfval);
                nelems = ARR_DIMS(arr)[0];
                elems = (int16 *) ARR_DATA_PTR(arr);
 
                for (natt = 0; natt < nelems; natt++)
                    included_cols = bms_add_member(included_cols, elems[natt]);
            }
 
            ExecClearTuple(tslot);
        }
        ExecDropSingleTupleTableSlot(tslot);
 
        walrcv_clear_result(pubres);
    }
 
    /*
     * Now fetch column names and types.
     */
    resetStringInfo(&cmd);
    appendStringInfoString(&cmd,
                           "SELECT a.attnum,"
                           "       a.attname,"
                           "       a.atttypid,"
                           "       a.attnum = ANY(i.indkey)");
 
    /* Generated columns can be replicated since version 18. */
    if (server_version >= 180000)
        appendStringInfoString(&cmd, ", a.attgenerated != ''");
 
    appendStringInfo(&cmd,
                     "  FROM pg_catalog.pg_attribute a"
                     "  LEFT JOIN pg_catalog.pg_index i"
                     "       ON (i.indexrelid = pg_get_replica_identity_index(%u))"
                     " WHERE a.attnum > 0::pg_catalog.int2"
                     "   AND NOT a.attisdropped %s"
                     "   AND a.attrelid = %u"
                     " ORDER BY a.attnum",
                     lrel->remoteid,
                     (server_version >= 120000 && server_version < 180000 ?
                      "AND a.attgenerated = ''" : ""),
                     lrel->remoteid);
    res = walrcv_exec(LogRepWorkerWalRcvConn, cmd.data,
                      server_version >= 180000 ? lengthof(attrRow) : lengthof(attrRow) - 1, attrRow);
 
    if (res->status != WALRCV_OK_TUPLES)
        ereport(ERROR,
                (errcode(ERRCODE_CONNECTION_FAILURE),
                 errmsg("could not fetch table info for table \"%s.%s\" from publisher: %s",
                        nspname, relname, res->err)));
 
    /* We don't know the number of rows coming, so allocate enough space. */
    lrel->attnames = palloc0(MaxTupleAttributeNumber * sizeof(char *));
    lrel->atttyps = palloc0(MaxTupleAttributeNumber * sizeof(Oid));
    lrel->attkeys = NULL;
 
    /*
     * Store the columns as a list of names.  Ignore those that are not
     * present in the column list, if there is one.
     */
    natt = 0;
    slot = MakeSingleTupleTableSlot(res->tupledesc, &TTSOpsMinimalTuple);
    while (tuplestore_gettupleslot(res->tuplestore, true, false, slot))
    {
        char       *rel_colname;
        AttrNumber  attnum;
 
        attnum = DatumGetInt16(slot_getattr(slot, 1, &isnull));
        Assert(!isnull);
 
        /* If the column is not in the column list, skip it. */
        if (included_cols != NULL && !bms_is_member(attnum, included_cols))
        {
            ExecClearTuple(slot);
            continue;
        }
 
        rel_colname = TextDatumGetCString(slot_getattr(slot, 2, &isnull));
        Assert(!isnull);
 
        lrel->attnames[natt] = rel_colname;
        lrel->atttyps[natt] = DatumGetObjectId(slot_getattr(slot, 3, &isnull));
        Assert(!isnull);
 
        if (DatumGetBool(slot_getattr(slot, 4, &isnull)))
            lrel->attkeys = bms_add_member(lrel->attkeys, natt);
 
        /* Remember if the remote table has published any generated column. */
        if (server_version >= 180000 && !(*gencol_published))
        {
            *gencol_published = DatumGetBool(slot_getattr(slot, 5, &isnull));
            Assert(!isnull);
        }
 
        /* Should never happen. */
        if (++natt >= MaxTupleAttributeNumber)
            elog(ERROR, "too many columns in remote table \"%s.%s\"",
                 nspname, relname);
 
        ExecClearTuple(slot);
    }
    ExecDropSingleTupleTableSlot(slot);
 
    lrel->natts = natt;
 
    walrcv_clear_result(res);
 
    /*
     * Get relation's row filter expressions. DISTINCT avoids the same
     * expression of a table in multiple publications from being included
     * multiple times in the final expression.
     *
     * We need to copy the row even if it matches just one of the
     * publications, so we later combine all the quals with OR.
     *
     * For initial synchronization, row filtering can be ignored in following
     * cases:
     *
     * 1) one of the subscribed publications for the table hasn't specified
     * any row filter
     *
     * 2) one of the subscribed publications has puballtables set to true
     *
     * 3) one of the subscribed publications is declared as TABLES IN SCHEMA
     * that includes this relation
     */
    if (server_version >= 150000)
    {
        /* Reuse the already-built pub_names. */
        Assert(pub_names != NULL);
 
        /* Check for row filters. */
        resetStringInfo(&cmd);
        appendStringInfo(&cmd,
                         "SELECT DISTINCT pg_get_expr(gpt.qual, gpt.relid)"
                         "  FROM pg_publication p,"
                         "  LATERAL pg_get_publication_tables(p.pubname) gpt"
                         " WHERE gpt.relid = %u"
                         "   AND p.pubname IN ( %s )",
                         lrel->remoteid,
                         pub_names->data);
 
        res = walrcv_exec(LogRepWorkerWalRcvConn, cmd.data, 1, qualRow);
 
        if (res->status != WALRCV_OK_TUPLES)
            ereport(ERROR,
                    (errmsg("could not fetch table WHERE clause info for table \"%s.%s\" from publisher: %s",
                            nspname, relname, res->err)));
 
        /*
         * Multiple row filter expressions for the same table will be combined
         * by COPY using OR. If any of the filter expressions for this table
         * are null, it means the whole table will be copied. In this case it
         * is not necessary to construct a unified row filter expression at
         * all.
         */
        slot = MakeSingleTupleTableSlot(res->tupledesc, &TTSOpsMinimalTuple);
        while (tuplestore_gettupleslot(res->tuplestore, true, false, slot))
        {
            Datum       rf = slot_getattr(slot, 1, &isnull);
 
            if (!isnull)
                *qual = lappend(*qual, makeString(TextDatumGetCString(rf)));
            else
            {
                /* Ignore filters and cleanup as necessary. */
                if (*qual)
                {
                    list_free_deep(*qual);
                    *qual = NIL;
                }
                break;
            }
 
            ExecClearTuple(slot);
        }
        ExecDropSingleTupleTableSlot(slot);
 
        walrcv_clear_result(res);
        destroyStringInfo(pub_names);
    }
 
    pfree(cmd.data);
}

References appendStringInfo(), appendStringInfoString(), ARR_DATA_PTR, ARR_DIMS, Assert, attnum, bms_add_member(), bms_is_member(), StringInfoData::data, DatumGetArrayTypeP, DatumGetBool(), DatumGetChar(), DatumGetInt16(), DatumGetObjectId(), destroyStringInfo(), elog, ereport, WalRcvExecResult::err, errcode(), errmsg(), ERROR, ExecClearTuple(), ExecDropSingleTupleTableSlot(), fb(), GetPublicationsStr(), initStringInfo(), lappend(), lengthof, list_free_deep(), LogRepWorkerWalRcvConn, MakeSingleTupleTableSlot(), makeString(), makeStringInfo(), MaxTupleAttributeNumber, MySubscription, NIL, palloc0(), pfree(), Subscription::publications, quote_literal_cstr(), relname, resetStringInfo(), server_version, slot_getattr(), WalRcvExecResult::status, TextDatumGetCString, TTSOpsMinimalTuple, WalRcvExecResult::tupledesc, WalRcvExecResult::tuplestore, tuplestore_gettupleslot(), tuplestore_tuple_count(), walrcv_clear_result(), walrcv_exec, WALRCV_OK_TUPLES, and walrcv_server_version.

Referenced by copy_table().

HasSubscriptionTablesCached()

bool HasSubscriptionTablesCached

(

void

)

Definition at line 1628 of file tablesync.c.

{
    bool        started_tx;
    bool        has_tables;
 
    /* We need up-to-date subscription tables info here */
    FetchRelationStates(&has_tables, NULL, &started_tx);
 
    if (started_tx)
    {
        CommitTransactionCommand();
        pgstat_report_stat(true);
    }
 
    return has_tables;
}

References CommitTransactionCommand(), fb(), FetchRelationStates(), and pgstat_report_stat().

Referenced by wait_for_local_flush().

LogicalRepSyncTableStart()

static char * LogicalRepSyncTableStart ( XLogRecPtr *  origin_startpos )

static

Definition at line 1219 of file tablesync.c.

{
    char       *slotname;
    char       *err;
    char        relstate;
    XLogRecPtr  relstate_lsn;
    Relation    rel;
    AclResult   aclresult;
    WalRcvExecResult *res;
    char        originname[NAMEDATALEN];
    RepOriginId originid;
    UserContext ucxt;
    bool        must_use_password;
    bool        run_as_owner;
 
    /* Check the state of the table synchronization. */
    StartTransactionCommand();
    relstate = GetSubscriptionRelState(MyLogicalRepWorker->subid,
                                       MyLogicalRepWorker->relid,
                                       &relstate_lsn);
    CommitTransactionCommand();
 
    /* Is the use of a password mandatory? */
    must_use_password = MySubscription->passwordrequired &&
        !MySubscription->ownersuperuser;
 
    SpinLockAcquire(&MyLogicalRepWorker->relmutex);
    MyLogicalRepWorker->relstate = relstate;
    MyLogicalRepWorker->relstate_lsn = relstate_lsn;
    SpinLockRelease(&MyLogicalRepWorker->relmutex);
 
    /*
     * If synchronization is already done or no longer necessary, exit now
     * that we've updated shared memory state.
     */
    switch (relstate)
    {
        case SUBREL_STATE_SYNCDONE:
        case SUBREL_STATE_READY:
        case SUBREL_STATE_UNKNOWN:
            FinishSyncWorker(); /* doesn't return */
    }
 
    /* Calculate the name of the tablesync slot. */
    slotname = (char *) palloc(NAMEDATALEN);
    ReplicationSlotNameForTablesync(MySubscription->oid,
                                    MyLogicalRepWorker->relid,
                                    slotname,
                                    NAMEDATALEN);
 
    /*
     * Here we use the slot name instead of the subscription name as the
     * application_name, so that it is different from the leader apply worker,
     * so that synchronous replication can distinguish them.
     */
    LogRepWorkerWalRcvConn =
        walrcv_connect(MySubscription->conninfo, true, true,
                       must_use_password,
                       slotname, &err);
    if (LogRepWorkerWalRcvConn == NULL)
        ereport(ERROR,
                (errcode(ERRCODE_CONNECTION_FAILURE),
                 errmsg("table synchronization worker for subscription \"%s\" could not connect to the publisher: %s",
                        MySubscription->name, err)));
 
    Assert(MyLogicalRepWorker->relstate == SUBREL_STATE_INIT ||
           MyLogicalRepWorker->relstate == SUBREL_STATE_DATASYNC ||
           MyLogicalRepWorker->relstate == SUBREL_STATE_FINISHEDCOPY);
 
    /* Assign the origin tracking record name. */
    ReplicationOriginNameForLogicalRep(MySubscription->oid,
                                       MyLogicalRepWorker->relid,
                                       originname,
                                       sizeof(originname));
 
    if (MyLogicalRepWorker->relstate == SUBREL_STATE_DATASYNC)
    {
        /*
         * We have previously errored out before finishing the copy so the
         * replication slot might exist. We want to remove the slot if it
         * already exists and proceed.
         *
         * XXX We could also instead try to drop the slot, last time we failed
         * but for that, we might need to clean up the copy state as it might
         * be in the middle of fetching the rows. Also, if there is a network
         * breakdown then it wouldn't have succeeded so trying it next time
         * seems like a better bet.
         */
        ReplicationSlotDropAtPubNode(LogRepWorkerWalRcvConn, slotname, true);
    }
    else if (MyLogicalRepWorker->relstate == SUBREL_STATE_FINISHEDCOPY)
    {
        /*
         * The COPY phase was previously done, but tablesync then crashed
         * before it was able to finish normally.
         */
        StartTransactionCommand();
 
        /*
         * The origin tracking name must already exist. It was created first
         * time this tablesync was launched.
         */
        originid = replorigin_by_name(originname, false);
        replorigin_session_setup(originid, 0);
        replorigin_session_origin = originid;
        *origin_startpos = replorigin_session_get_progress(false);
 
        CommitTransactionCommand();
 
        goto copy_table_done;
    }
 
    SpinLockAcquire(&MyLogicalRepWorker->relmutex);
    MyLogicalRepWorker->relstate = SUBREL_STATE_DATASYNC;
    MyLogicalRepWorker->relstate_lsn = InvalidXLogRecPtr;
    SpinLockRelease(&MyLogicalRepWorker->relmutex);
 
    /*
     * Update the state, create the replication origin, and make them visible
     * to others.
     */
    StartTransactionCommand();
    UpdateSubscriptionRelState(MyLogicalRepWorker->subid,
                               MyLogicalRepWorker->relid,
                               MyLogicalRepWorker->relstate,
                               MyLogicalRepWorker->relstate_lsn,
                               false);
 
    /*
     * Create the replication origin in a separate transaction from the one
     * that sets up the origin in shared memory. This prevents the risk that
     * changes to the origin in shared memory cannot be rolled back if the
     * transaction aborts.
     */
    originid = replorigin_by_name(originname, true);
    if (!OidIsValid(originid))
        originid = replorigin_create(originname);
 
    CommitTransactionCommand();
    pgstat_report_stat(true);
 
    StartTransactionCommand();
 
    /*
     * Use a standard write lock here. It might be better to disallow access
     * to the table while it's being synchronized. But we don't want to block
     * the main apply process from working and it has to open the relation in
     * RowExclusiveLock when remapping remote relation id to local one.
     */
    rel = table_open(MyLogicalRepWorker->relid, RowExclusiveLock);
 
    /*
     * Start a transaction in the remote node in REPEATABLE READ mode.  This
     * ensures that both the replication slot we create (see below) and the
     * COPY are consistent with each other.
     */
    res = walrcv_exec(LogRepWorkerWalRcvConn,
                      "BEGIN READ ONLY ISOLATION LEVEL REPEATABLE READ",
                      0, NULL);
    if (res->status != WALRCV_OK_COMMAND)
        ereport(ERROR,
                (errcode(ERRCODE_CONNECTION_FAILURE),
                 errmsg("table copy could not start transaction on publisher: %s",
                        res->err)));
    walrcv_clear_result(res);
 
    /*
     * Create a new permanent logical decoding slot. This slot will be used
     * for the catchup phase after COPY is done, so tell it to use the
     * snapshot to make the final data consistent.
     */
    walrcv_create_slot(LogRepWorkerWalRcvConn,
                       slotname, false /* permanent */ , false /* two_phase */ ,
                       MySubscription->failover,
                       CRS_USE_SNAPSHOT, origin_startpos);
 
    /*
     * Advance the origin to the LSN got from walrcv_create_slot and then set
     * up the origin. The advancement is WAL logged for the purpose of
     * recovery. Locks are to prevent the replication origin from vanishing
     * while advancing.
     *
     * The purpose of doing these before the copy is to avoid doing the copy
     * again due to any error in advancing or setting up origin tracking.
     */
    LockRelationOid(ReplicationOriginRelationId, RowExclusiveLock);
    replorigin_advance(originid, *origin_startpos, InvalidXLogRecPtr,
                       true /* go backward */ , true /* WAL log */ );
    UnlockRelationOid(ReplicationOriginRelationId, RowExclusiveLock);
 
    replorigin_session_setup(originid, 0);
    replorigin_session_origin = originid;
 
    /*
     * If the user did not opt to run as the owner of the subscription
     * ('run_as_owner'), then copy the table as the owner of the table.
     */
    run_as_owner = MySubscription->runasowner;
    if (!run_as_owner)
        SwitchToUntrustedUser(rel->rd_rel->relowner, &ucxt);
 
    /*
     * Check that our table sync worker has permission to insert into the
     * target table.
     */
    aclresult = pg_class_aclcheck(RelationGetRelid(rel), GetUserId(),
                                  ACL_INSERT);
    if (aclresult != ACLCHECK_OK)
        aclcheck_error(aclresult,
                       get_relkind_objtype(rel->rd_rel->relkind),
                       RelationGetRelationName(rel));
 
    /*
     * COPY FROM does not honor RLS policies.  That is not a problem for
     * subscriptions owned by roles with BYPASSRLS privilege (or superuser,
     * who has it implicitly), but other roles should not be able to
     * circumvent RLS.  Disallow logical replication into RLS enabled
     * relations for such roles.
     */
    if (check_enable_rls(RelationGetRelid(rel), InvalidOid, false) == RLS_ENABLED)
        ereport(ERROR,
                (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
                 errmsg("user \"%s\" cannot replicate into relation with row-level security enabled: \"%s\"",
                        GetUserNameFromId(GetUserId(), true),
                        RelationGetRelationName(rel))));
 
    /* Now do the initial data copy */
    PushActiveSnapshot(GetTransactionSnapshot());
    copy_table(rel);
    PopActiveSnapshot();
 
    res = walrcv_exec(LogRepWorkerWalRcvConn, "COMMIT", 0, NULL);
    if (res->status != WALRCV_OK_COMMAND)
        ereport(ERROR,
                (errcode(ERRCODE_CONNECTION_FAILURE),
                 errmsg("table copy could not finish transaction on publisher: %s",
                        res->err)));
    walrcv_clear_result(res);
 
    if (!run_as_owner)
        RestoreUserContext(&ucxt);
 
    table_close(rel, NoLock);
 
    /* Make the copy visible. */
    CommandCounterIncrement();
 
    /*
     * Update the persisted state to indicate the COPY phase is done; make it
     * visible to others.
     */
    UpdateSubscriptionRelState(MyLogicalRepWorker->subid,
                               MyLogicalRepWorker->relid,
                               SUBREL_STATE_FINISHEDCOPY,
                               MyLogicalRepWorker->relstate_lsn,
                               false);
 
    CommitTransactionCommand();
 
copy_table_done:
 
    elog(DEBUG1,
         "LogicalRepSyncTableStart: '%s' origin_startpos lsn %X/%08X",
         originname, LSN_FORMAT_ARGS(*origin_startpos));
 
    /*
     * We are done with the initial data synchronization, update the state.
     */
    SpinLockAcquire(&MyLogicalRepWorker->relmutex);
    MyLogicalRepWorker->relstate = SUBREL_STATE_SYNCWAIT;
    MyLogicalRepWorker->relstate_lsn = *origin_startpos;
    SpinLockRelease(&MyLogicalRepWorker->relmutex);
 
    /*
     * Finally, wait until the leader apply worker tells us to catch up and
     * then return to let LogicalRepApplyLoop do it.
     */
    wait_for_worker_state_change(SUBREL_STATE_CATCHUP);
    return slotname;
}

References ACL_INSERT, aclcheck_error(), ACLCHECK_OK, Assert, check_enable_rls(), CommandCounterIncrement(), CommitTransactionCommand(), Subscription::conninfo, copy_table(), CRS_USE_SNAPSHOT, DEBUG1, elog, ereport, WalRcvExecResult::err, err(), errcode(), errmsg(), ERROR, Subscription::failover, fb(), FinishSyncWorker(), get_relkind_objtype(), GetSubscriptionRelState(), GetTransactionSnapshot(), GetUserId(), GetUserNameFromId(), InvalidOid, InvalidXLogRecPtr, LockRelationOid(), LogRepWorkerWalRcvConn, LSN_FORMAT_ARGS, MyLogicalRepWorker, MySubscription, Subscription::name, NAMEDATALEN, NoLock, Subscription::oid, OidIsValid, Subscription::ownersuperuser, palloc(), Subscription::passwordrequired, pg_class_aclcheck(), pgstat_report_stat(), PopActiveSnapshot(), PushActiveSnapshot(), RelationData::rd_rel, RelationGetRelationName, RelationGetRelid, LogicalRepWorker::relid, LogicalRepWorker::relmutex, LogicalRepWorker::relstate, LogicalRepWorker::relstate_lsn, ReplicationOriginNameForLogicalRep(), ReplicationSlotDropAtPubNode(), ReplicationSlotNameForTablesync(), replorigin_advance(), replorigin_by_name(), replorigin_create(), replorigin_session_get_progress(), replorigin_session_origin, replorigin_session_setup(), RestoreUserContext(), RLS_ENABLED, RowExclusiveLock, Subscription::runasowner, SpinLockAcquire, SpinLockRelease, StartTransactionCommand(), WalRcvExecResult::status, LogicalRepWorker::subid, SwitchToUntrustedUser(), table_close(), table_open(), UnlockRelationOid(), UpdateSubscriptionRelState(), wait_for_worker_state_change(), walrcv_clear_result(), walrcv_connect, walrcv_create_slot, walrcv_exec, and WALRCV_OK_COMMAND.

Referenced by start_table_sync().

make_copy_attnamelist()

static List * make_copy_attnamelist ( LogicalRepRelMapEntry *  rel )

static

Definition at line 626 of file tablesync.c.

{
    List       *attnamelist = NIL;
    int         i;
 
    for (i = 0; i < rel->remoterel.natts; i++)
    {
        attnamelist = lappend(attnamelist,
                              makeString(rel->remoterel.attnames[i]));
    }
 
 
    return attnamelist;
}

References LogicalRepRelation::attnames, fb(), i, lappend(), makeString(), LogicalRepRelation::natts, NIL, and LogicalRepRelMapEntry::remoterel.

Referenced by copy_table().

ProcessSyncingTablesForApply()

void ProcessSyncingTablesForApply

(

XLogRecPtr

current_lsn

)

Definition at line 368 of file tablesync.c.

{
    struct tablesync_start_time_mapping
    {
        Oid         relid;
        TimestampTz last_start_time;
    };
    static HTAB *last_start_times = NULL;
    ListCell   *lc;
    bool        started_tx;
    bool        should_exit = false;
    Relation    rel = NULL;
 
    Assert(!IsTransactionState());
 
    /* We need up-to-date sync state info for subscription tables here. */
    FetchRelationStates(NULL, NULL, &started_tx);
 
    /*
     * Prepare a hash table for tracking last start times of workers, to avoid
     * immediate restarts.  We don't need it if there are no tables that need
     * syncing.
     */
    if (table_states_not_ready != NIL && !last_start_times)
    {
        HASHCTL     ctl;
 
        ctl.keysize = sizeof(Oid);
        ctl.entrysize = sizeof(struct tablesync_start_time_mapping);
        last_start_times = hash_create("Logical replication table sync worker start times",
                                       256, &ctl, HASH_ELEM | HASH_BLOBS);
    }
 
    /*
     * Clean up the hash table when we're done with all tables (just to
     * release the bit of memory).
     */
    else if (table_states_not_ready == NIL && last_start_times)
    {
        hash_destroy(last_start_times);
        last_start_times = NULL;
    }
 
    /*
     * Process all tables that are being synchronized.
     */
    foreach(lc, table_states_not_ready)
    {
        SubscriptionRelState *rstate = (SubscriptionRelState *) lfirst(lc);
 
        if (!started_tx)
        {
            StartTransactionCommand();
            started_tx = true;
        }
 
        Assert(get_rel_relkind(rstate->relid) != RELKIND_SEQUENCE);
 
        if (rstate->state == SUBREL_STATE_SYNCDONE)
        {
            /*
             * Apply has caught up to the position where the table sync has
             * finished.  Mark the table as ready so that the apply will just
             * continue to replicate it normally.
             */
            if (current_lsn >= rstate->lsn)
            {
                char        originname[NAMEDATALEN];
 
                rstate->state = SUBREL_STATE_READY;
                rstate->lsn = current_lsn;
 
                /*
                 * Remove the tablesync origin tracking if exists.
                 *
                 * There is a chance that the user is concurrently performing
                 * refresh for the subscription where we remove the table
                 * state and its origin or the tablesync worker would have
                 * already removed this origin. We can't rely on tablesync
                 * worker to remove the origin tracking as if there is any
                 * error while dropping we won't restart it to drop the
                 * origin. So passing missing_ok = true.
                 *
                 * Lock the subscription and origin in the same order as we
                 * are doing during DDL commands to avoid deadlocks. See
                 * AlterSubscription_refresh.
                 */
                LockSharedObject(SubscriptionRelationId, MyLogicalRepWorker->subid,
                                 0, AccessShareLock);
 
                if (!rel)
                    rel = table_open(SubscriptionRelRelationId, RowExclusiveLock);
 
                ReplicationOriginNameForLogicalRep(MyLogicalRepWorker->subid,
                                                   rstate->relid,
                                                   originname,
                                                   sizeof(originname));
                replorigin_drop_by_name(originname, true, false);
 
                /*
                 * Update the state to READY only after the origin cleanup.
                 */
                UpdateSubscriptionRelState(MyLogicalRepWorker->subid,
                                           rstate->relid, rstate->state,
                                           rstate->lsn, true);
            }
        }
        else
        {
            LogicalRepWorker *syncworker;
 
            /*
             * Look for a sync worker for this relation.
             */
            LWLockAcquire(LogicalRepWorkerLock, LW_SHARED);
 
            syncworker = logicalrep_worker_find(WORKERTYPE_TABLESYNC,
                                                MyLogicalRepWorker->subid,
                                                rstate->relid, false);
 
            if (syncworker)
            {
                /* Found one, update our copy of its state */
                SpinLockAcquire(&syncworker->relmutex);
                rstate->state = syncworker->relstate;
                rstate->lsn = syncworker->relstate_lsn;
                if (rstate->state == SUBREL_STATE_SYNCWAIT)
                {
                    /*
                     * Sync worker is waiting for apply.  Tell sync worker it
                     * can catchup now.
                     */
                    syncworker->relstate = SUBREL_STATE_CATCHUP;
                    syncworker->relstate_lsn =
                        Max(syncworker->relstate_lsn, current_lsn);
                }
                SpinLockRelease(&syncworker->relmutex);
 
                /* If we told worker to catch up, wait for it. */
                if (rstate->state == SUBREL_STATE_SYNCWAIT)
                {
                    /* Signal the sync worker, as it may be waiting for us. */
                    if (syncworker->proc)
                        logicalrep_worker_wakeup_ptr(syncworker);
 
                    /* Now safe to release the LWLock */
                    LWLockRelease(LogicalRepWorkerLock);
 
                    if (started_tx)
                    {
                        /*
                         * We must commit the existing transaction to release
                         * the existing locks before entering a busy loop.
                         * This is required to avoid any undetected deadlocks
                         * due to any existing lock as deadlock detector won't
                         * be able to detect the waits on the latch.
                         *
                         * Also close any tables prior to the commit.
                         */
                        if (rel)
                        {
                            table_close(rel, NoLock);
                            rel = NULL;
                        }
                        CommitTransactionCommand();
                        pgstat_report_stat(false);
                    }
 
                    /*
                     * Enter busy loop and wait for synchronization worker to
                     * reach expected state (or die trying).
                     */
                    StartTransactionCommand();
                    started_tx = true;
 
                    wait_for_table_state_change(rstate->relid,
                                                SUBREL_STATE_SYNCDONE);
                }
                else
                    LWLockRelease(LogicalRepWorkerLock);
            }
            else
            {
                /*
                 * If there is no sync worker for this table yet, count
                 * running sync workers for this subscription, while we have
                 * the lock.
                 */
                int         nsyncworkers =
                    logicalrep_sync_worker_count(MyLogicalRepWorker->subid);
                struct tablesync_start_time_mapping *hentry;
                bool        found;
 
                /* Now safe to release the LWLock */
                LWLockRelease(LogicalRepWorkerLock);
 
                hentry = hash_search(last_start_times, &rstate->relid,
                                     HASH_ENTER, &found);
                if (!found)
                    hentry->last_start_time = 0;
 
                launch_sync_worker(WORKERTYPE_TABLESYNC, nsyncworkers,
                                   rstate->relid, &hentry->last_start_time);
            }
        }
    }
 
    /* Close table if opened */
    if (rel)
        table_close(rel, NoLock);
 
 
    if (started_tx)
    {
        /*
         * Even when the two_phase mode is requested by the user, it remains
         * as 'pending' until all tablesyncs have reached READY state.
         *
         * When this happens, we restart the apply worker and (if the
         * conditions are still ok) then the two_phase tri-state will become
         * 'enabled' at that time.
         *
         * Note: If the subscription has no tables then leave the state as
         * PENDING, which allows ALTER SUBSCRIPTION ... REFRESH PUBLICATION to
         * work.
         */
        if (MySubscription->twophasestate == LOGICALREP_TWOPHASE_STATE_PENDING)
        {
            CommandCounterIncrement();  /* make updates visible */
            if (AllTablesyncsReady())
            {
                ereport(LOG,
                        (errmsg("logical replication apply worker for subscription \"%s\" will restart so that two_phase can be enabled",
                                MySubscription->name)));
                should_exit = true;
            }
        }
 
        CommitTransactionCommand();
        pgstat_report_stat(true);
    }
 
    if (should_exit)
    {
        /*
         * Reset the last-start time for this worker so that the launcher will
         * restart it without waiting for wal_retrieve_retry_interval.
         */
        ApplyLauncherForgetWorkerStartTime(MySubscription->oid);
 
        proc_exit(0);
    }
}

References AccessShareLock, AllTablesyncsReady(), ApplyLauncherForgetWorkerStartTime(), Assert, CommandCounterIncrement(), CommitTransactionCommand(), ctl, ereport, errmsg(), fb(), FetchRelationStates(), get_rel_relkind(), HASH_BLOBS, hash_create(), hash_destroy(), HASH_ELEM, HASH_ENTER, hash_search(), IsTransactionState(), HASHCTL::keysize, last_start_times, launch_sync_worker(), lfirst, LockSharedObject(), LOG, logicalrep_sync_worker_count(), logicalrep_worker_find(), logicalrep_worker_wakeup_ptr(), SubscriptionRelState::lsn, LW_SHARED, LWLockAcquire(), LWLockRelease(), Max, MyLogicalRepWorker, MySubscription, Subscription::name, NAMEDATALEN, NIL, NoLock, Subscription::oid, pgstat_report_stat(), proc_exit(), SubscriptionRelState::relid, ReplicationOriginNameForLogicalRep(), replorigin_drop_by_name(), RowExclusiveLock, SpinLockAcquire, SpinLockRelease, StartTransactionCommand(), SubscriptionRelState::state, LogicalRepWorker::subid, table_close(), table_open(), table_states_not_ready, Subscription::twophasestate, UpdateSubscriptionRelState(), wait_for_table_state_change(), and WORKERTYPE_TABLESYNC.

Referenced by ProcessSyncingRelations().

ProcessSyncingTablesForSync()

void ProcessSyncingTablesForSync

(

XLogRecPtr

current_lsn

)

Definition at line 244 of file tablesync.c.

{
    SpinLockAcquire(&MyLogicalRepWorker->relmutex);
 
    if (MyLogicalRepWorker->relstate == SUBREL_STATE_CATCHUP &&
        current_lsn >= MyLogicalRepWorker->relstate_lsn)
    {
        TimeLineID  tli;
        char        syncslotname[NAMEDATALEN] = {0};
        char        originname[NAMEDATALEN] = {0};
 
        MyLogicalRepWorker->relstate = SUBREL_STATE_SYNCDONE;
        MyLogicalRepWorker->relstate_lsn = current_lsn;
 
        SpinLockRelease(&MyLogicalRepWorker->relmutex);
 
        /*
         * UpdateSubscriptionRelState must be called within a transaction.
         */
        if (!IsTransactionState())
            StartTransactionCommand();
 
        UpdateSubscriptionRelState(MyLogicalRepWorker->subid,
                                   MyLogicalRepWorker->relid,
                                   MyLogicalRepWorker->relstate,
                                   MyLogicalRepWorker->relstate_lsn,
                                   false);
 
        /*
         * End streaming so that LogRepWorkerWalRcvConn can be used to drop
         * the slot.
         */
        walrcv_endstreaming(LogRepWorkerWalRcvConn, &tli);
 
        /*
         * Cleanup the tablesync slot.
         *
         * This has to be done after updating the state because otherwise if
         * there is an error while doing the database operations we won't be
         * able to rollback dropped slot.
         */
        ReplicationSlotNameForTablesync(MyLogicalRepWorker->subid,
                                        MyLogicalRepWorker->relid,
                                        syncslotname,
                                        sizeof(syncslotname));
 
        /*
         * It is important to give an error if we are unable to drop the slot,
         * otherwise, it won't be dropped till the corresponding subscription
         * is dropped. So passing missing_ok = false.
         */
        ReplicationSlotDropAtPubNode(LogRepWorkerWalRcvConn, syncslotname, false);
 
        CommitTransactionCommand();
        pgstat_report_stat(false);
 
        /*
         * Start a new transaction to clean up the tablesync origin tracking.
         * This transaction will be ended within the FinishSyncWorker(). Now,
         * even, if we fail to remove this here, the apply worker will ensure
         * to clean it up afterward.
         *
         * We need to do this after the table state is set to SYNCDONE.
         * Otherwise, if an error occurs while performing the database
         * operation, the worker will be restarted and the in-memory state of
         * replication progress (remote_lsn) won't be rolled-back which would
         * have been cleared before restart. So, the restarted worker will use
         * invalid replication progress state resulting in replay of
         * transactions that have already been applied.
         */
        StartTransactionCommand();
 
        ReplicationOriginNameForLogicalRep(MyLogicalRepWorker->subid,
                                           MyLogicalRepWorker->relid,
                                           originname,
                                           sizeof(originname));
 
        /*
         * Resetting the origin session removes the ownership of the slot.
         * This is needed to allow the origin to be dropped.
         */
        replorigin_session_reset();
        replorigin_session_origin = InvalidRepOriginId;
        replorigin_session_origin_lsn = InvalidXLogRecPtr;
        replorigin_session_origin_timestamp = 0;
 
        /*
         * Drop the tablesync's origin tracking if exists.
         *
         * There is a chance that the user is concurrently performing refresh
         * for the subscription where we remove the table state and its origin
         * or the apply worker would have removed this origin. So passing
         * missing_ok = true.
         */
        replorigin_drop_by_name(originname, true, false);
 
        FinishSyncWorker();
    }
    else
        SpinLockRelease(&MyLogicalRepWorker->relmutex);
}

References CommitTransactionCommand(), fb(), FinishSyncWorker(), InvalidRepOriginId, InvalidXLogRecPtr, IsTransactionState(), LogRepWorkerWalRcvConn, MyLogicalRepWorker, NAMEDATALEN, pgstat_report_stat(), LogicalRepWorker::relid, LogicalRepWorker::relmutex, LogicalRepWorker::relstate, LogicalRepWorker::relstate_lsn, ReplicationOriginNameForLogicalRep(), ReplicationSlotDropAtPubNode(), ReplicationSlotNameForTablesync(), replorigin_drop_by_name(), replorigin_session_origin, replorigin_session_origin_lsn, replorigin_session_origin_timestamp, replorigin_session_reset(), SpinLockAcquire, SpinLockRelease, StartTransactionCommand(), LogicalRepWorker::subid, UpdateSubscriptionRelState(), and walrcv_endstreaming.

Referenced by ProcessSyncingRelations().

ReplicationSlotNameForTablesync()

void ReplicationSlotNameForTablesync	(	Oid	suboid,
		Oid	relid,
		char *	syncslotname,
		Size	szslot
	)

Definition at line 1203 of file tablesync.c.

{
    snprintf(syncslotname, szslot, "pg_%u_sync_%u_" UINT64_FORMAT, suboid,
             relid, GetSystemIdentifier());
}

References fb(), GetSystemIdentifier(), snprintf, and UINT64_FORMAT.

Referenced by AlterSubscription_refresh(), DropSubscription(), LogicalRepSyncTableStart(), ProcessSyncingTablesForSync(), and ReportSlotConnectionError().

run_tablesync_worker()

static void run_tablesync_worker ( void  )

static

Definition at line 1552 of file tablesync.c.

{
    char        originname[NAMEDATALEN];
    XLogRecPtr  origin_startpos = InvalidXLogRecPtr;
    char       *slotname = NULL;
    WalRcvStreamOptions options;
 
    start_table_sync(&origin_startpos, &slotname);
 
    ReplicationOriginNameForLogicalRep(MySubscription->oid,
                                       MyLogicalRepWorker->relid,
                                       originname,
                                       sizeof(originname));
 
    set_apply_error_context_origin(originname);
 
    set_stream_options(&options, slotname, &origin_startpos);
 
    walrcv_startstreaming(LogRepWorkerWalRcvConn, &options);
 
    /* Apply the changes till we catchup with the apply worker. */
    start_apply(origin_startpos);
}

References fb(), InvalidXLogRecPtr, LogRepWorkerWalRcvConn, MyLogicalRepWorker, MySubscription, NAMEDATALEN, Subscription::oid, LogicalRepWorker::relid, ReplicationOriginNameForLogicalRep(), set_apply_error_context_origin(), set_stream_options(), start_apply(), start_table_sync(), and walrcv_startstreaming.

Referenced by TableSyncWorkerMain().

start_table_sync()

static void start_table_sync ( XLogRecPtr *  origin_startpos, char **  slotname  )

static

Definition at line 1509 of file tablesync.c.

{
    char       *sync_slotname = NULL;
 
    Assert(am_tablesync_worker());
 
    PG_TRY();
    {
        /* Call initial sync. */
        sync_slotname = LogicalRepSyncTableStart(origin_startpos);
    }
    PG_CATCH();
    {
        if (MySubscription->disableonerr)
            DisableSubscriptionAndExit();
        else
        {
            /*
             * Report the worker failed during table synchronization. Abort
             * the current transaction so that the stats message is sent in an
             * idle state.
             */
            AbortOutOfAnyTransaction();
            pgstat_report_subscription_error(MySubscription->oid,
                                             WORKERTYPE_TABLESYNC);
 
            PG_RE_THROW();
        }
    }
    PG_END_TRY();
 
    /* allocate slot name in long-lived context */
    *slotname = MemoryContextStrdup(ApplyContext, sync_slotname);
    pfree(sync_slotname);
}

References AbortOutOfAnyTransaction(), am_tablesync_worker(), ApplyContext, Assert, Subscription::disableonerr, DisableSubscriptionAndExit(), fb(), LogicalRepSyncTableStart(), MemoryContextStrdup(), MySubscription, Subscription::oid, pfree(), PG_CATCH, PG_END_TRY, PG_RE_THROW, PG_TRY, pgstat_report_subscription_error(), and WORKERTYPE_TABLESYNC.

Referenced by run_tablesync_worker().

TableSyncWorkerMain()

void TableSyncWorkerMain

(

Datum

main_arg

)

Definition at line 1578 of file tablesync.c.

{
    int         worker_slot = DatumGetInt32(main_arg);
 
    SetupApplyOrSyncWorker(worker_slot);
 
    run_tablesync_worker();
 
    FinishSyncWorker();
}

References DatumGetInt32(), fb(), FinishSyncWorker(), run_tablesync_worker(), and SetupApplyOrSyncWorker().

UpdateTwoPhaseState()

void UpdateTwoPhaseState	(	Oid	suboid,
		char	new_state
	)

Definition at line 1649 of file tablesync.c.

{
    Relation    rel;
    HeapTuple   tup;
    bool        nulls[Natts_pg_subscription];
    bool        replaces[Natts_pg_subscription];
    Datum       values[Natts_pg_subscription];
 
    Assert(new_state == LOGICALREP_TWOPHASE_STATE_DISABLED ||
           new_state == LOGICALREP_TWOPHASE_STATE_PENDING ||
           new_state == LOGICALREP_TWOPHASE_STATE_ENABLED);
 
    rel = table_open(SubscriptionRelationId, RowExclusiveLock);
    tup = SearchSysCacheCopy1(SUBSCRIPTIONOID, ObjectIdGetDatum(suboid));
    if (!HeapTupleIsValid(tup))
        elog(ERROR,
             "cache lookup failed for subscription oid %u",
             suboid);
 
    /* Form a new tuple. */
    memset(values, 0, sizeof(values));
    memset(nulls, false, sizeof(nulls));
    memset(replaces, false, sizeof(replaces));
 
    /* And update/set two_phase state */
    values[Anum_pg_subscription_subtwophasestate - 1] = CharGetDatum(new_state);
    replaces[Anum_pg_subscription_subtwophasestate - 1] = true;
 
    tup = heap_modify_tuple(tup, RelationGetDescr(rel),
                            values, nulls, replaces);
    CatalogTupleUpdate(rel, &tup->t_self, tup);
 
    heap_freetuple(tup);
    table_close(rel, RowExclusiveLock);
}

References Assert, CatalogTupleUpdate(), CharGetDatum(), elog, ERROR, fb(), heap_freetuple(), heap_modify_tuple(), HeapTupleIsValid, ObjectIdGetDatum(), RelationGetDescr, RowExclusiveLock, SearchSysCacheCopy1, table_close(), table_open(), and values.

Referenced by CreateSubscription(), and run_apply_worker().

wait_for_table_state_change()

static bool wait_for_table_state_change ( Oid  relid, char  expected_state  )

static

Definition at line 140 of file tablesync.c.

{
    char        state;
 
    for (;;)
    {
        LogicalRepWorker *worker;
        XLogRecPtr  statelsn;
 
        CHECK_FOR_INTERRUPTS();
 
        InvalidateCatalogSnapshot();
        state = GetSubscriptionRelState(MyLogicalRepWorker->subid,
                                        relid, &statelsn);
 
        if (state == SUBREL_STATE_UNKNOWN)
            break;
 
        if (state == expected_state)
            return true;
 
        /* Check if the sync worker is still running and bail if not. */
        LWLockAcquire(LogicalRepWorkerLock, LW_SHARED);
        worker = logicalrep_worker_find(WORKERTYPE_TABLESYNC,
                                        MyLogicalRepWorker->subid, relid,
                                        false);
        LWLockRelease(LogicalRepWorkerLock);
        if (!worker)
            break;
 
        (void) WaitLatch(MyLatch,
                         WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH,
                         1000L, WAIT_EVENT_LOGICAL_SYNC_STATE_CHANGE);
 
        ResetLatch(MyLatch);
    }
 
    return false;
}

References CHECK_FOR_INTERRUPTS, fb(), GetSubscriptionRelState(), InvalidateCatalogSnapshot(), logicalrep_worker_find(), LW_SHARED, LWLockAcquire(), LWLockRelease(), MyLatch, MyLogicalRepWorker, ResetLatch(), LogicalRepWorker::subid, WaitLatch(), WL_EXIT_ON_PM_DEATH, WL_LATCH_SET, WL_TIMEOUT, and WORKERTYPE_TABLESYNC.

Referenced by ProcessSyncingTablesForApply().

wait_for_worker_state_change()

static bool wait_for_worker_state_change ( char  expected_state )

static

Definition at line 189 of file tablesync.c.

{
    int         rc;
 
    for (;;)
    {
        LogicalRepWorker *worker;
 
        CHECK_FOR_INTERRUPTS();
 
        /*
         * Done if already in correct state.  (We assume this fetch is atomic
         * enough to not give a misleading answer if we do it with no lock.)
         */
        if (MyLogicalRepWorker->relstate == expected_state)
            return true;
 
        /*
         * Bail out if the apply worker has died, else signal it we're
         * waiting.
         */
        LWLockAcquire(LogicalRepWorkerLock, LW_SHARED);
        worker = logicalrep_worker_find(WORKERTYPE_APPLY,
                                        MyLogicalRepWorker->subid, InvalidOid,
                                        false);
        if (worker && worker->proc)
            logicalrep_worker_wakeup_ptr(worker);
        LWLockRelease(LogicalRepWorkerLock);
        if (!worker)
            break;
 
        /*
         * Wait.  We expect to get a latch signal back from the apply worker,
         * but use a timeout in case it dies without sending one.
         */
        rc = WaitLatch(MyLatch,
                       WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH,
                       1000L, WAIT_EVENT_LOGICAL_SYNC_STATE_CHANGE);
 
        if (rc & WL_LATCH_SET)
            ResetLatch(MyLatch);
    }
 
    return false;
}

References CHECK_FOR_INTERRUPTS, fb(), InvalidOid, logicalrep_worker_find(), logicalrep_worker_wakeup_ptr(), LW_SHARED, LWLockAcquire(), LWLockRelease(), MyLatch, MyLogicalRepWorker, LogicalRepWorker::proc, LogicalRepWorker::relstate, ResetLatch(), LogicalRepWorker::subid, WaitLatch(), WL_EXIT_ON_PM_DEATH, WL_LATCH_SET, WL_TIMEOUT, and WORKERTYPE_APPLY.

Referenced by LogicalRepSyncTableStart().

Variable Documentation

copybuf

StringInfo copybuf = NULL

static

Definition at line 127 of file tablesync.c.

Referenced by copy_read_data(), copy_table(), CopyStreamReceive(), dumpTableData_copy(), GetCopyDataByte(), GetCopyDataEnd(), GetCopyDataString(), GetCopyDataUInt64(), HandleCopyStream(), ProcessKeepaliveMsg(), ProcessWALDataMsg(), ReceiveArchiveStreamChunk(), ReceiveBackupManifestChunk(), ReceiveBackupManifestInMemoryChunk(), ReceiveCopyData(), ReceiveTarCopyChunk(), ReportCopyDataParseError(), and StreamLogicalLog().

table_states_not_ready

List* table_states_not_ready = NIL

Definition at line 125 of file tablesync.c.

Referenced by AllTablesyncsReady(), FetchRelationStates(), and ProcessSyncingTablesForApply().