On Thu, Apr 2, 2015 at 9:23 AM, Pietro Pugni wrote:
> the command
> dmesg | grep -i numa
> doesn’t display me anything. I think T420 hasn’t NUMA on it. Is there a
> way to enable it from Ubuntu? I don’t have immediate access to BIOS (server
> is in another location).
>
NUMA stands for "Non-Unif
On Thu, Apr 2, 2015 at 6:33 AM, Pietro Pugni wrote:
> *T420*
> work_mem = 512MB
>
> *MacMini*
> work_mem = 32MB
>
So that is why the T420 does memory sorts and the mini does disk sorts.
I'd start looking at why memory sorts on the T420 is so slow. Check your
numa settings, etc (as already
On Wed, Apr 1, 2015 at 9:56 AM, Pietro Pugni wrote:
> *Now let’s propose some query profiling times.*
>
> B type set are transactions, so it's impossible for me to post EXPLAIN
> ANALYZE results. I've extracted two querys from a single transactions and
> executed the twos on both system. Here ar
to write a WAL record, seeks somewhere else to read something, etc.
Or, if you put WAL on a journalled FS, even if it's on dedicated spindles ;-)
a.
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca
an return...
a.
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca command like a king,
http://www.highrise.ca/ work like a slave.
--
Sent via pgsql-performance mail
0% utilized?
a.
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca command like a king,
http://www.highrise.ca/ work like a slave.
--
Sent via pgsql-performance mailing
ns that may be *very* fruitful.
1) Multicolumn index (depending on insert/update/delete patterns)
2) partition by date (depending on query types)
3) rollup views of history (depending on query types)
4) trigger based mat-view style rollups (depending on
insert/update/delete
to the point of migrating large XFS filesystems to
ext4 for production databases yet? Or at least using ext4 in new
large-scale filesystems for PG?
a.
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca command
al throughput...
Of course, linux has completely re-written/changed the
sync/barrier/flush methods over the past few years, and there is no
guarantee they don't keep changing the implementation details in the
future, so keep up on the filesystem details of whatever you&#
ow a wrench into the plan, what happens when during normal
battery tests, your raid controller decides the battery is failing...
of course, it's going to start screaming and send all your monitoring
alarms off (you're monitoring that, right?), but have you thought to
make sure that your FS i
isk and only
find out that it was in the disks (small) cache, and you're raid is
out of sync after an outage because of that... I believe most raid
cards will handle that correctly for you automatically.
a.
--
Aidan Van Dyk Create like a god,
ai.
und it uses a "local"
postgres (server and psql) using the source default of /tmp?
a.
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca command like a king,
http://www.highrise.ca/
required for the kernel to setup the mappings is
less than the repeated syscalls of a simple write().
All those things seem like something that somebody could synthetically
benchmark to prove value before even trying to bolt into PostgreSQL.
a.
--
Aidan Van Dyk
th acceptable responses since
the fsync/barreir hasn't "completed".
Or have I missed another subtle interaction?
a.
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca comma
on a RAID5, even with BBU is worse than WAL on a
dedicated set of RAID1 spindles with it's own BBU.
Wow!
;-)
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca command like a king,
h
* Greg Smith [100121 09:49]:
> Aidan Van Dyk wrote:
>> Sure, if your WAL is on the same FS as your data, you're going to get
>> hit, and *especially* on ext3...
>>
>> But, I think that's one of the reasons people usually recommend putting
>> WA
d putting
WAL separate. Even if it's just another partition on the same (set of)
disk(s), you get the benefit of not having to wait for all the dirty
ext3 pages from your whole database FS to be flushed before the WAL write
can complete on it's own FS.
a.
--
Aidan Van Dyk
aditional blocks being that are either read or
written to disk are going to completely chop your 43MB/s rate...
a.
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca command like a king,
http://www.highri
this again
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca command like a king,
http://www.highrise.ca/ work like a slave.
signature.asc
Description: Digital signature
; the OS and the disks. If you access a table via index scan though, all its
> pages will go through shared_buffers.
In older version too, or only since synchronized scans got in?
a.
--
Aidan Van Dyk Create like
't let me have it as a dedicated DB-only machine. Go
> figure :) doing other stuff the cache tends to turn over pretty quickly.
All the more reason to find a way to use it all as shared buffers and
lock it into ram...
Oh, sorry, you expect the DB to play nice with everything els
tails matter, and results trump theory, so test it ;-)
a.
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca command like a king,
http://www.highrise.ca/ work like a slave.
signature.asc
Description: Digital signature
re cases).
The other thing that scares me about raid-5 is the write-hole, and the
possible delayed inconsistency that brings...
Again, hopefully mitigated by a dependable controller w/ BBU...
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca
The
forking isn't the expensive part, the per-database setup that happens is
the expensive setup... All pre-forking would save us is a tiny part of
the initial setup, and in turn make our robust postmaster controller no
longer have control.
a.
-
throw a new execution of the query to
PostgreSQL, you get to just do #3, the easy quick part, especially for
small simple queries where all the data is in shared buffers or the cache.
a.
--
Aidan Van Dyk Create like a god,
ai...@highrise.ca
little money, and get good "read only"
performance ...
... as long as you don't ever have a disk or system crash ...
... or can afford to rebuild if you do ...
... etc ...
--
Aidan Van Dyk Create like a god,
[EMAIL PROTECTED]
int of asking about it's poor performance is...
a.
--
Aidan Van Dyk Create like a god,
[EMAIL PROTECTED] command like a king,
http://www.highrise.ca/ work like a slave.
s
* Joshua D. Drake <[EMAIL PROTECTED]> [081209 11:01]:
> Yes the SmartArray series is quite common and actually know to perform
> reasonably well, in RAID 10. You still appear to be trying RAID 5.
*boggle*
Are people *still* using raid5?
/me gives up!
--
A
Seems like this didn't make it through to the list the first time...
* Aidan Van Dyk <[EMAIL PROTECTED]> [081106 22:19]:
> * David Rees <[EMAIL PROTECTED]> [081106 21:22]:
>
> > 2. A write barrier instructs the lower level hardware that commands
> > issued be
y MVCC safe. From the TODO:
* CLUSTER
o Make CLUSTER preserve recently-dead tuples per MVCC
requirements
But the documents don't mention anything about cluster being unsafe.
AFAIK, Vacuum full doesn't suffer the same MVCC issues that cluster
does. Is th
30 matches
Mail list logo
