FailedAssertion("!(PrivateRefCount[i] == 0)", File: "bufmgr.c", Line: 1741

pg 9.2 git master
AMD 8120 (8-core) / 6 GB memory / Centos 6.2


I have experimented a bit with dropping a table from master, then querying that table from a
sync-rep slave.  It is a little worrying that this, the first test I tried, crashes the slave.

There are two instance on one machine, head1 (=master) and head2 (=sync-rep slave).

First, I generated a tab-separated file, a one off, to be used in the test:

echo "
    copy (
        select
              repeat('X',20)        as c1
            , repeat('X',20)        as c2
            , repeat('X',20)        as c3
            , repeat('X',20)        as c4
            , repeat('X',20)        as c5
        from generate_series(1, 200000)
    )
    to stdout
    csv delimiter E'\t';
" |  $HOME/pg_stuff/pg_installations/pgsql.head1/bin/psql -p 6564 -d testdb > dropload_copy.txt


That txt file is zipped, and the actual test consists of a bash while loop which
  1. drops the table
  2. loads the file into the table
  3. Either:
          a.  nothing
          b.  does a select count(*) on the table


So, it repeats the following:

zcat dropload_copy.txt.gz \
 | grep -v '^#' \
 | $HOME/pg_stuff/pg_installations/pgsql.head1/bin/psql -p 6564 -d testdb -c "
drop table if exists t;
create table t (
        c1 text,
        c2 text,
        c3 text,
        c4 text,
        c5 text
);
copy t from stdin csv delimiter E'\t';
analyze t;";

PAUSE_DURATION=0
PSQL=$HOME/pg_stuff/pg_installations/pgsql.head1/bin/psql

if [[ 0 -eq 1 ]];  #  ON-OFF switch
then

    echo "sleep $PAUSE_DURATION"
          sleep $PAUSE_DURATION;
    (
        echo "select current_setting('port') port, count(*) from $schema.$table" | $PSQL -qtXp 6564 -d
testdb  # master
        echo "select current_setting('port') port, count(*) from $schema.$table" | $PSQL -qtXp 6565 -d
testdb  # wal_receiver_01
    #echo "select current_setting('port') port, count(*) from $schema.$table" | $PSQL -qtXp 6566 -d
testdb  # wal_receiver_02
    ) | grep -v '^$'
fi



This runs fine for hours on end, as long as the ON-OFF switch is disabled.

But when that if-block is added the client crashes after a while (sometimes almost immediately; it
never survives longer then 20 minutes):

2012-05-26 10:44:22.617 CEST 10274 ERROR:  could not fsync file "base/21268/32807": No such file
or directory
2012-05-26 10:44:28.465 CEST 10274 ERROR:  could not fsync file "base/21268/32867": No such file
or directory
2012-05-26 10:44:28.587 CEST 10270 FATAL:  could not open file "base/21268/32994": No such file or
directory
2012-05-26 10:44:28.588 CEST 10270 CONTEXT:  writing block 2508 of relation base/21268/32994
        xlog redo multi-insert (init): rel 1663/21268/33006; blk 3117; 58 tuples
TRAP: FailedAssertion("!(PrivateRefCount[i] == 0)", File: "bufmgr.c", Line: 1741)
2012-05-26 10:44:31.131 CEST 10269 LOG:  startup process (PID 10270) was terminated by signal 6:
Aborted
2012-05-26 10:44:31.131 CEST 10269 LOG:  terminating any other active server processes


Crazy scenario , I'll admit, but surely this shouldn't be able to crash the client?

I attach the logfiles of master(=head1) and slave (=head2).  It show how the above ran for an hour
without problems (while the ON/OFF switch was disabled), and how the crash came quickly when I
switched it on (to add the select count(*) statements).


Erik Rijkers