We now have production running on 10.6 (including all those extensions), and have seen no more issues at all. So I am assuming it must be fixed in 10.6 at this point.
If it's not gone with 10.6, could you print the two variables in that expression?
Well it is gone with 10.6 I think, but here is what you requested on my 10.5 version:
(gdb) frame 2
#2 0x0000000000a45f9c in ExceptionalCondition (conditionName=0xc2d688 "!(prev_first_lsn < cur_txn->first_lsn)", errorType=0xc2d404 "FailedAssertion", fileName=0xc2d478 "reorderbuffer.c", lineNumber=688) at assert.c:54
54abort();
(gdb) up
#3 0x000000000084b0ac in AssertTXNLsnOrder (rb=0x2775790) at reorderbuffer.c:688
688Assert(prev_first_lsn < cur_txn->first_lsn);
(gdb) print prev_first_lsn
$2 = 9888781386112
(gdb) print cur_txn->first_lsn
$3 = 9888781386112
> # Install Extensions > cd $HOME/pgl_ddl_deploy > make clean > sudo "PATH=$PATH" make install > cd $HOME/pglogical_ticker > make clean > sudo "PATH=$PATH" make install > cd $HOME/pg_fact_loader > make clean > sudo "PATH=$PATH" make install
It's entirely possible that one of those extensions does something to violate WAL logging rules :/
But I would expect then to still have a problem in our production instance running 10.6, which I don't.
