Re: LDC - Load Distributed Checkpoints with PG8.3b2 on Solaris

I was waiting to digest what I saw before sending it to the group

I am running EAStress workload

I am using odata_sync which should sync as soon as it is written

with checkpoint_completion_target=0.9 and checkpoint_time=5m it seems to
be doing the right thing from the logfile output

2007-11-13 09:20:49.070 PST 9180 LOG: checkpoint starting: time
2007-11-13 09:21:13.808 PST 9458 LOG: automatic analyze of table
"specdb.public.o_orderline" system usage: CPU 0.03s/0.50u sec elapsed
7.79 sec
2007-11-13 09:21:19.830 PST 9458 LOG: automatic vacuum of table
"specdb.public.txn_log_table": index scans: 1
pages: 11 removed, 105 remain
tuples: 3147 removed, 40 remain
system usage: CPU 0.11s/0.09u sec elapsed 6.02 sec
2007-11-13 09:22:12.112 PST 9462 LOG: automatic vacuum of table
"specdb.public.txn_log_table": index scans: 1
pages: 28 removed, 77 remain
tuples: 1990 removed, 95 remain
system usage: CPU 0.11s/0.09u sec elapsed 5.98 sec
2007-11-13 09:23:12.121 PST 9466 LOG: automatic vacuum of table
"specdb.public.txn_log_table": index scans: 1
pages: 0 removed, 77 remain
tuples: 3178 removed, 128 remain
system usage: CPU 0.11s/0.04u sec elapsed 5.87 sec
2007-11-13 09:24:12.220 PST 9470 LOG: automatic vacuum of table
"specdb.public.txn_log_table": index scans: 1
pages: 0 removed, 77 remain
tuples: 3394 removed, 57 remain
system usage: CPU 0.11s/0.04u sec elapsed 5.85 sec
2007-11-13 09:25:12.400 PST 9474 LOG: automatic vacuum of table
"specdb.public.txn_log_table": index scans: 1
pages: 0 removed, 77 remain
tuples: 3137 removed, 1 remain
system usage: CPU 0.11s/0.04u sec elapsed 5.93 sec
2007-11-13 09:25:18.723 PST 9180 LOG: checkpoint complete: wrote 33362
buffers (2.2%); 0 transaction log file(s) added, 0 removed, 0 recycled;
write=269.642 s, sync=0.003 s, total=269.653 s
2007-11-13 09:25:49.000 PST 9180 LOG: checkpoint starting: time

However actual iostat output still shows non-uniform distribution but I
havent put the exact time stamp on the iostat outputs to correlate that
with the logfile entries.. Maybe I should do that.

So from the PostgreSQL view things are doing fine based on outputs: I
need to figure out the Solaris view on it now.

Could it be related to autovacuum happening also?

Regards,
Jignesh

Tom Lane wrote:
> "Jignesh K. Shah" <J(dot)K(dot)Shah(at)Sun(dot)COM> writes:
>
>> I will turn on checkpoint_logging to get more idea as Heikki suggested
>>
>
> Did you find out anything?
>
> Did this happen on every checkpoint, or only some of them? The bug
> Itagaki-san pointed out today in IsCheckpointOnSchedule might account
> for some checkpoints being done at full speed, but not all ...
>
> regards, tom lane
>