like:
backend_hostname0 = 'host1'
backend_port0 = 5432
backend_weight0 = 1
backend_data_directory0 = '/var/lib/postgresql/8.3/main'
backend_hostname1 = 'host2'
backend_port1 = 5433
backend_weight1 = 1
backend_data_directory1 = '/var/lib/postgresql/8.3/main'
--
Toma
OVERY_DATA
For example, it includes $DATA variable, which is not used anywhere.
Is it the same as "/data" referenced later? Probably yes.
Anyway, I can't get it to work.
Does anyone have online recover with pgpool-ii explained in more details?
--
Tomasz Chmielewski
http://w
Jaume Sabater wrote:
On Mon, Oct 26, 2009 at 5:59 PM, Tomasz Chmielewski wrote:
Is it the same as "/data" referenced later? Probably yes.
It is.
Anyway, I can't get it to work.
It proved very useful to me in order to create mine, but I did not
expect it to work out
Jaume Sabater wrote:
On Mon, Oct 26, 2009 at 10:32 PM, Tomasz Chmielewski wrote:
Yes, this is what I don't understand - these scripts seem to want to copy
the whole postgresql data directory. That can't be right, can it?
If you use rsync, only the differences.
Assuming your data
Jaume Sabater wrote:
On Mon, Oct 26, 2009 at 5:59 PM, Tomasz Chmielewski wrote:
Is it the same as "/data" referenced later? Probably yes.
It is.
Anyway, I can't get it to work.
It proved very useful to me in order to create mine, but I did not
expect it to work out
Tomasz Chmielewski wrote:
Anyway, I can't get it to work.
It proved very useful to me in order to create mine, but I did not
expect it to work out of the box. Nor should you. It's meant to be an
example for you to learn, just that. It's a script that works on
Tatsuo's l
If pgpool-ii is restarted, it will consider that all backends are in
"good" state - even if one of the backends was not up to date (and data
was not being replicated there).
What is the best way to preserve backend state across pgpool-ii restarts?
--
Tomasz Chmielewski
http:
hing which could cause it would be OOM-killer).
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
the nodes is out of sync. And later,
finding the node which was out of sync would be real pain.
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
Tomasz Chmielewski wrote:
When one node is gone, pgpool-ii can execute a "failover_command".
After a node is recovered, pgpool-ii can execute a "failback_command".
What I lack (or can't find) is something executed between those two
commands: a command executed
s reachable again).
Is something like this possible with pgpool-ii?
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
_hostname0 = 'local'
backend_weight0 = 0
backend_hostname1 = 'remote'
backend_weight1 = 1
Or maybe:
backend_hostname0 = 'local'
backend_weight0 = 1
backend_hostname1 = 'remote'
backend_weight1 = 0
Or is the mean
gres2'
backend_weight1 = 1
backend_hostname1 = 'postgres3'
backend_weight1 = 2
postgres1 should not get any reads.
postgres2 should get 33,3(3)% reads.
postgres2 should get 66,6(6)% reads.
Is my reasoning correct?
--
Tomasz Chmielewski
http://wpkg.org
___
Tomasz Chmielewski wrote:
Thanks.
Out of curiosity, how would be such values handled?
I made some typos wrt numbers -once again, corrected below:
backend_hostname0 = 'postgres1'
backend_weight0 = 0
backend_hostname1 = 'postgres2'
backend_weight1 = 1
backend_ho
s.
postgres2 should get 33,3(3)% reads.
postgres2 should get 66,6(6)% reads.
Is my reasoning correct?
Yes. See select_load_balancing_node() in child.c for more details.
OK, thanks for your help.
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-gener
guess status of down node is what the NEWS file talks about?
- pgpool_status file is created under logdir. So you need to
give write permission to the directory so that pgpool-II can
read/write pgpool_status file.
--
Tomasz Chmielewski
http://wpkg.org
Is there any "torture test" for pgpool-ii?
Say, a test which would connect over pgpool-ii and do the following:
- create a database,
- do lots of inserts etc. changes, over a certain period of time (30
mins or so),
- ideally, compare databases which pgpool-ii uses.
--
Tomasz C
On 13.12.2009 02:03, Tomasz Chmielewski wrote:
> Is there any "torture test" for pgpool-ii?
>
> Say, a test which would connect over pgpool-ii and do the following:
>
> - create a database,
>
> - do lots of inserts etc. changes, over a certain period of time (30
On 13.12.2009 02:57, Tomasz Chmielewski wrote:
What's wrong here? I use pgpool-2.3.
If I do this right before rsyncing on the "to-be-recovered" server, then
I don't get the errors anymore:
/usr/bin/pg_ctlcluster 8.3 main stop --force
I followed this guide, which
ained, nor outlined in example scripts.
See this thread from yesterday:
"torture test" for pgpool-ii?
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
ived: host=[local]
Dec 18 12:06:49 pg1 pgpool: 2009-12-18 12:06:49 LOG: pid 4352: connection
received: host=[local]
It doesn't look too healthy to me - what exactly happened there?
This is a setup of two postgres servers, connections go through pgpool-ii 2.
husagestats.id"
2009-12-18 01:39:42 UTC NOTICE: CREATE TABLE / PRIMARY KEY will create
implicit index "core_bandwidthusagestats_pkey" for table
"core_bandwidthusagestats"
until:
2009-12-18 01:39:42 UTC NOTICE: CREATE TABLE / UNIQUE will create implicit
index "auth_user_user_permissions_user_id_key" for table
"auth_user_user_permissions"
2009-12-18 01:39:49 UTC ERROR: null value in column "object_id" violates
not-null constraint
2009-12-18 01:39:49 UTC STATEMENT: INSERT INTO "core_event"
("id","ip","when","bytes","protocol_id","country") VALUES
(800,E'serv1',E'2009-12-09 12:00:13',3
2009-12-18 01:39:49 UTC ERROR: current transaction is aborted, commands
ignored until end of transaction block
2009-12-18 01:39:49 UTC STATEMENT: INSERT INTO "django_content_type"
("id","name","app_label","model") VALUES
(1,E'permission',E'auth',E'permission');
2009-12-18 01:39:49 UTC ERROR: current transaction is aborted, commands
ignored until end of transaction block
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
2009-12-23 06:03:20 SGT LOG: autovacuum launcher started
2009-12-23 06:03:20 SGT LOG: database system is ready to accept connections
Because of these errors, recovery sometimes fails.
How does postgres on the node which is recovered determines the %f files it
needs
On 22.12.2009 23:05, Tomasz Chmielewski wrote:
> I followed the http://linuxsilo.net/articles/postgresql-pgpool.html to set up
> pgpool-ii replication.
>
>
> When I detach and recover a node with these commands:
>
> # pcp_detach_node -d 240 127.0.0.1 9898 user pass 1
>
DEFAULT
[393-1] 2010-03-01 03:06:56 SGT LOG: duration: 0.028 ms statement: COMMIT
And they are repeated every second or so.
Does anyone have an idea why pgpool queries one of the backends so often?
The queries I made by pgpool, as they stop when I stop pgpool.
I use pgpool-2.3.2.2, but I had thi
query.
It may well be that the queries are sent by my application running in
the network - but:
- backend0 has weight 1; backend1 has weight 0 - doesn't in mean these
queries should be sent to backend0? Right now, backend0 gets only few
queries; backend1 gets thousands.
--
Tomasz Chmiel
Tomasz Chmielewski wrote:
> - backend0 has weight 1; backend1 has weight 0 - doesn't in mean these
> queries should be sent to backend0? Right now, backend0 gets only few
> queries; backend1 gets thousands.
I guess I didn't reload postgres settings on one of the servers, so
turn the slave into Master and
have it replicate everything to a new DB server when it is brought online?
heartbeat?
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman
elopped
my solution to check if backends are synchronized or not. Do you think
that could interest people? Or do you know an existing tool?
Surely that would be interesting.
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpo
op it because I need it.
Before publishing it, I have to resolve some formal legal issues with
one of my clients. I'll tell you when it is published.
Just pinging here - was it forgotten, or are there any legal issues here
etc.?
--
Tomasz Chmiele
y to kill pgpool in this state is to use kill -9.
Is it a known issue?
I use pgpool-II 2.3.2.2.
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
at it fails when we start pcp_attch_node one after another?
2. Is there a way to attach multiple nodes in an "atomic" way?
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
which go to node 0 only
7. yo do "pcp_attach_node ... 0" to attach node 1
8. writes go to both nodes now, but node 1 does not contain some data now
There is a workaround: stop pgpool, remove its status file, start it again, but
that's not real
grams (such as pgadmin) which
maintain a persistent connection and prevent pgpool from performing
recovery.
Just out of curiosity, why don't you just stop pgadmin if you go into
online recovery?
There can be clients connected you have no control over.
--
Tomasz Chmielewski
http:
t;netstat -tpna" to see the ports processes are listening on.
Also, "man ps" and "man netstat" could be interesting to you.
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http
LOG: pid 22202: pgpool-II
successfully started. version 3.0.3 (umiyameboshi)
>From what I found, there was indeed such an issue before 3.0.1. Did I miss
>anything?
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing lis
On 31.03.2011 17:31, Tomasz Chmielewski wrote:
> I'm just trying to upgrade from 3.0.1 to 3.0.3.
>
> Unfortunately, it complains like below:
>
> Mar 31 23:16:09 db20 pgpool: 2011-03-31 23:16:09 ERROR: pid 22202:
> pool_init_pool_passwd: couldn't open /etc/poo
from frontend. frontend abnormally
exited
I get them in my syslog around 40 times per hour.
Is that anything good? Anything bad? Neutral?
Other than that, everything else seem to be working correctly.
--
Tomasz Chmielewski
http://wpkg.org
1493 live rows and 0
dead rows; 1493
May 12 05:23:04 db20 postgres[27590]: [73-2] rows in sample, 1362565 estimated
total rows
May 12 05:23:04 db20 postgres[27590]: [74-1] 2011-05-12 05:23:04 SGT LOG:
duration: 1126957.051 ms statement: VACUUM VERBOSE ANALYZE core_extevent;
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
On 12.05.2011 00:21, Tomasz Chmielewski wrote:
> I removed a large number of records from a table (~38 mln out of ~40 mln),
> after which I've started "VACUUM VERBOSE ANALYZE my_table".
>
> Near the very end, pgpool kicked one of the backends - is it expected? Pl
you.
Are there any other commands which should not be executed through
pgpool? For example, is REINDEX TABLE safe to execute through pgpool?
Probably I should add this to FAQ.
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing l
t all, I should expect them on both backends, or?
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
ncy among db nodes
server closed the connection unexpectedly
This probably means the server terminated abnormally
before or while processing the request.
connection to server was lost
So after all, looks REINDEX is also not so safe?
--
Tomasz Chmielewski
http://wpkg.org
h pgpool? As I
understand, it does not change database data visible to the clients, so
should be safe.
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
rect me if I'm wrong.
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
ld be done during the least load, but in theory, a
deadlock can still happen.
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
during the least load, but in theory, a
deadlock can still happen.
Why don't you use an explicit transatcion plus lock to avoid deadlock
for the first place?
BEGIN;
LOCK TABLE
REINDEX TABLE
END;
That could be helpful, thank you.
--
Tomasz Chmielewski
http://wpk
, you will need to remove this file.
--
Tomasz Chmielewski
http://wpkg.org
___
Pgpool-general mailing list
Pgpool-general@pgfoundry.org
http://pgfoundry.org/mailman/listinfo/pgpool-general
ation which tries to connect/write all the time.
If pgpool starts and the application connects/writes before you start
pcp_attach_node, you will have problems (even if time before starting
pgpool and pcp_attach_node may seem short, the application may still be
able to write to pgpool).
--
T
49 matches
Mail list logo