postgres-xc-general Mailing List for Postgres-XC (Page 2)

If my memory is correct, you can issue DML only once in plpgsql functions.

Regards;
---
Koichi Suzuki

2014/06/12 0:15、Jan Keirse <jan...@tv...> のメール：

> Hello all,
> 
> I'm currently evaluating if we could use postgres-xc to make some
> databases scale beyond one machine.
> However I would like to combine sharding across hosts (distribute by
> hash as provided by postgres-xc) with standard postgresql trigger
> based partitioning.
> The release notes for postgres xc claim that DML cannot be used in
> plpgsql functions (here:
> http://postgres-xc.sourceforge.net/docs/1_2_beta/release-xc-1-2.html )
> I'm trying to understand exactly what this statement means. My
> expectation was that an insert statement inside a plpgsql function
> would somehow raise an error or fail.
> 
> However, when I have an insert trigger on a table like this:
> 
> CREATE TRIGGER partition_trg
>  BEFORE INSERT
>  ON history
>  FOR EACH ROW
>  EXECUTE PROCEDURE trg_partition('day');
> 
> and the trg_partition function looks like this:
> 
> --------------------------------------------------------------------------------------------------------------------------------------
> CREATE OR REPLACE FUNCTION trg_partition()
>  RETURNS trigger AS
> $BODY$
> 
> DECLARE
> prefix text := 'partitions.';
> timeformat text;
> selector text;
> _interval INTERVAL;
> tablename text;
> startdate text;
> enddate text;
> create_table_part text;
> create_index_part text;
> BEGIN
> 
> selector = TG_ARGV[0];
> 
> IF selector = 'day' THEN
> timeformat := 'YYYY_MM_DD';
> ELSIF selector = 'month' THEN
> timeformat := 'YYYY_MM';
> END IF;
> 
> _interval := '1 ' || selector;
> tablename :=  TG_TABLE_NAME || '_p' || TO_CHAR(TO_TIMESTAMP(NEW.clock)
> at time zone 'gmt', timeformat);
> 
> EXECUTE 'INSERT INTO ' || prefix || quote_ident(tablename) || ' SELECT
> ($1).*' USING NEW;
> RETURN NULL;
> 
> EXCEPTION
> WHEN undefined_table THEN
> 
> startdate := EXTRACT(epoch FROM date_trunc(selector,
> TO_TIMESTAMP(NEW.clock) at time zone 'gmt'));
> enddate := EXTRACT(epoch FROM date_trunc(selector,
> TO_TIMESTAMP(NEW.clock) at time zone 'gmt' + _interval ));
> 
> create_table_part:= 'CREATE TABLE '|| prefix || quote_ident(tablename)
> || ' (CHECK ((clock >= ' || quote_literal(startdate) || ' AND clock <
> ' || quote_literal(enddate) || '))) INHERITS ( public.'||
> TG_TABLE_NAME || ') DISTRIBUTE BY HASH(itemid)';
> create_index_part:= 'ALTER TABLE ' || prefix || quote_ident(tablename)
> || ' ADD CONSTRAINT ' || quote_ident(tablename) || '_pkey PRIMARY KEY
> (itemid,clock)';
> 
> EXECUTE create_table_part;
> EXECUTE create_index_part;
> 
> --insert it again
> EXECUTE 'INSERT INTO ' || prefix || quote_ident(tablename) || ' SELECT
> ($1).*' USING NEW;
> RETURN NULL;
> 
> END;
> 
> $BODY$
>  LANGUAGE plpgsql VOLATILE
>  COST 100;
> ALTER FUNCTION trg_partition()
>  OWNER TO postgres;
> 
> --------------------------------------------------------------------------------------------------------------------------------------
> 
> And I do an insert into history like this:
> 
> insert into history (itemid, clock, value) values ( 1,1,123);
> 
> 
> The partition history_p1970_01_01 does get created and the record is
> inserted fine and is visible accross all nodes coördinator nodes and
> in the expected data node.
> If I do a second insert into the table, resulting in an insert in the
> same partition, for example like this:
> 
> insert into history (itemid, clock, value) values ( 100,1,123);
> 
> The results are a bit surprising: at first it looks like nothing
> happened, the new record doesn't show up. I guess that's why it's not
> supported. However when I disconnect the session that did the insert,
> all of a sudden the data does become visible in the
> history_p1970_01_01 partition. It's as if the data doesn't get
> committed until the session is disconnected.
> 
> Should I throw the towel in the ring and give up on this, or is there
> an easy workaround for this so that DML can in fact be used with some
> reservations.
> 
> The only detail I could find on the subject was this post:
> http://sourceforge.net/p/postgres-xc/mailman/message/31433999/
> that appears to indicate that it could indeed probably work when
> taking some things into account.
> 
> On another note, does this limitation apply to all pl/... languages or
> only to pl/pgsql? If the cause of these problems is strictly related
> to pl/pgsql and would not apply to say plv8 I could rewrite my
> triggers.
> 
> Kind Regards,
> 
> Jan Keirse
> 
> -- 
> 
> 
> **** DISCLAIMER ****
> 
> http://www.tvh.com/newen2/emaildisclaimer/default.html
> 
> "This message is delivered to all addressees subject to the conditions
> set forth in the attached disclaimer, which is an integral part of this
> message."
> 
> ------------------------------------------------------------------------------
> HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions
> Find What Matters Most in Your Big Data with HPCC Systems
> Open Source. Fast. Scalable. Simple. Ideal for Dirty Data.
> Leverages Graph Analysis for Fast Processing & Easy Data Exploration
> http://p.sf.net/sfu/hpccsystems
> _______________________________________________
> Postgres-xc-general mailing list
> Pos...@li...
> https://lists.sourceforge.net/lists/listinfo/postgres-xc-general

Hello all,

I'm currently evaluating if we could use postgres-xc to make some
databases scale beyond one machine.
However I would like to combine sharding across hosts (distribute by
hash as provided by postgres-xc) with standard postgresql trigger
based partitioning.
The release notes for postgres xc claim that DML cannot be used in
plpgsql functions (here:
http://postgres-xc.sourceforge.net/docs/1_2_beta/release-xc-1-2.html )
I'm trying to understand exactly what this statement means. My
expectation was that an insert statement inside a plpgsql function
would somehow raise an error or fail.

However, when I have an insert trigger on a table like this:

CREATE TRIGGER partition_trg
  BEFORE INSERT
  ON history
  FOR EACH ROW
  EXECUTE PROCEDURE trg_partition('day');

and the trg_partition function looks like this:

--------------------------------------------------------------------------------------------------------------------------------------
CREATE OR REPLACE FUNCTION trg_partition()
  RETURNS trigger AS
$BODY$

DECLARE
prefix text := 'partitions.';
timeformat text;
selector text;
_interval INTERVAL;
tablename text;
startdate text;
enddate text;
create_table_part text;
create_index_part text;
BEGIN

selector = TG_ARGV[0];

IF selector = 'day' THEN
timeformat := 'YYYY_MM_DD';
ELSIF selector = 'month' THEN
timeformat := 'YYYY_MM';
END IF;

_interval := '1 ' || selector;
tablename :=  TG_TABLE_NAME || '_p' || TO_CHAR(TO_TIMESTAMP(NEW.clock)
at time zone 'gmt', timeformat);

EXECUTE 'INSERT INTO ' || prefix || quote_ident(tablename) || ' SELECT
($1).*' USING NEW;
RETURN NULL;

EXCEPTION
WHEN undefined_table THEN

startdate := EXTRACT(epoch FROM date_trunc(selector,
TO_TIMESTAMP(NEW.clock) at time zone 'gmt'));
enddate := EXTRACT(epoch FROM date_trunc(selector,
TO_TIMESTAMP(NEW.clock) at time zone 'gmt' + _interval ));

create_table_part:= 'CREATE TABLE '|| prefix || quote_ident(tablename)
|| ' (CHECK ((clock >= ' || quote_literal(startdate) || ' AND clock <
' || quote_literal(enddate) || '))) INHERITS ( public.'||
TG_TABLE_NAME || ') DISTRIBUTE BY HASH(itemid)';
create_index_part:= 'ALTER TABLE ' || prefix || quote_ident(tablename)
|| ' ADD CONSTRAINT ' || quote_ident(tablename) || '_pkey PRIMARY KEY
(itemid,clock)';

EXECUTE create_table_part;
EXECUTE create_index_part;

--insert it again
EXECUTE 'INSERT INTO ' || prefix || quote_ident(tablename) || ' SELECT
($1).*' USING NEW;
RETURN NULL;

END;

$BODY$
  LANGUAGE plpgsql VOLATILE
  COST 100;
ALTER FUNCTION trg_partition()
  OWNER TO postgres;

--------------------------------------------------------------------------------------------------------------------------------------

And I do an insert into history like this:

insert into history (itemid, clock, value) values ( 1,1,123);


The partition history_p1970_01_01 does get created and the record is
inserted fine and is visible accross all nodes coördinator nodes and
in the expected data node.
If I do a second insert into the table, resulting in an insert in the
same partition, for example like this:

insert into history (itemid, clock, value) values ( 100,1,123);

The results are a bit surprising: at first it looks like nothing
happened, the new record doesn't show up. I guess that's why it's not
supported. However when I disconnect the session that did the insert,
all of a sudden the data does become visible in the
history_p1970_01_01 partition. It's as if the data doesn't get
committed until the session is disconnected.

Should I throw the towel in the ring and give up on this, or is there
an easy workaround for this so that DML can in fact be used with some
reservations.

The only detail I could find on the subject was this post:
http://sourceforge.net/p/postgres-xc/mailman/message/31433999/
that appears to indicate that it could indeed probably work when
taking some things into account.

On another note, does this limitation apply to all pl/... languages or
only to pl/pgsql? If the cause of these problems is strictly related
to pl/pgsql and would not apply to say plv8 I could rewrite my
triggers.

Kind Regards,

Jan Keirse

-- 


**** DISCLAIMER ****

http://www.tvh.com/newen2/emaildisclaimer/default.html

"This message is delivered to all addressees subject to the conditions
set forth in the attached disclaimer, which is an integral part of this
message."

Good to hear that.

Please feel free to post your issue with pgxc_ctl as well.    Pgxc_ctl tutorial will be found at http://sourceforge.net/projects/postgres-xc/files/Pgxc_ctl_primer/

Hope it is much more helpful than reference document.

Best;
---
Koichi Suzuki

2014/06/11 2:37、Wilkerson, Daniel <dwi...@fu...> のメール：

> Thank you for the replies Masataka and Koichi. I appreciate them very much.
> 
> You were both right. It was definitely some config relates
> issues/mismatches. I suppose my mind was pretty tired after configuring
> each config file on everyone one of the servers and trying to troubleshoot
> all day yesterday. Cross connecting with psql and using netstat definitely
> helped suss out the problematic config items across the nodes and servers.
> 
> To fix my issue I made some corrections to pg_hba.conf for my
> coordinators. My data nodes were configured correctly, but the
> coordinators were not. I also noticed on CN1 and CN3, I missed changing
> the listen address from localhost to * (0.0.0.0). That issue was apparent
> with netstat -an
> 
> I'm going to have a look at the pgxc_ctl tool to simplify cluster starts
> and stops as well as the config of future clusters for production use.
> This is only a prototype environment.
> 
> Thanks again for the help my friends. I appreciate it.
> 
> dw
> 
> On 6/10/14 6:06 AM, "Masataka Saito" <pg...@gm...> wrote:
> 
>> Hello Wilkerson,
>> 
>> I'm afraid that you're using Postgres-XC 1.2 which has such ugly bug
>> fixed on 1.2.1.
>> Or Postgres-XC nodes must trust other nodes: you have to add
>> pg_hba.conf entry for subnet xx.xx.40.0/24 and xx.xx.33.0/24 with
>> trust authentication method, so you can't specify other authentication
>> method which requires extra dialogue with client, e.g. password or md5
>> are not allowed.
>> 
>> If both suggestions don't hit, please test that you can connect to
>> other nodes from cn1 by psql command. I think it's a good way to find
>> the component in which the cause resides.
>> 
>> Regards.
>> 
>> On 10 June 2014 07:13, Wilkerson, Daniel <dwi...@fu...> wrote:
>>> Hi. I am trying to stand up a multi node cluster across multiple
>>> servers. The current configuration I have is one coordinator and two
>>> datanodes per server on 3 servers. Each has a GTM proxy configured. I
>>> have two more servers configured to be the GTM master and slave/standby.
>>> 
>>> When executing create database in my cluster after all the config work
>>> (including the create node calls) I am getting the following error.
>>> ERROR:  Failed to get pooled connections
>>> CONTEXT:  SQL statement "EXECUTE DIRECT ON (cn2) 'SELECT
>>> pg_catalog.pg_try_advisory_xact_lock_shared(65535, 0)'"
>>> 
>>> I noticed that the coordinator on each node is added as host = local
>>> host after using the initdb command to create the coordinator nodes.
>>> When executing the create node calls to add the other coordinators and
>>> datanodes to each other I am using the IP addresses of the servers
>>> instead; I have also tried the DNS names also with no luck.
>>> 
>>> The output of pgxc_node from one of the cluster coordinator/datanodes
>>> looks like:
>>> 
>>> oid  | node_name | node_type | node_port |  node_host   |
>>> nodeis_primary | nodeis_preferred |   node_id
>>> 
>>> -------+-----------+-----------+-----------+--------------+--------------
>>> --+------------------+-------------
>>> 11198 | cn1       | C         |      5432 | localhost    | f
>>>  | f                | -1178713634
>>> 16393 | cn2       | C         |      5432 | xx.xx.40.11  | f
>>>  | f                | -1923125220
>>> 16394 | cn3       | C         |      5432 | xx.xx.40.123 | f
>>>  | f                |  1101067607
>>> 16395 | dn1       | D         |     15432 | xx.xx.33.198 | t
>>>  | t                |  -560021589
>>> 16396 | dn2       | D         |     15433 | xx.xx.33.198 | f
>>>  | f                |   352366662
>>> 16397 | dn3       | D         |     15432 | xx.xx.40.11  | f
>>>  | f                |  -700122826
>>> 16398 | dn4       | D         |     15433 | xx.xx.40.11  | f
>>>  | f                |   823103418
>>> 16399 | dn5       | D         |     15432 | xx.xx.40.123 | f
>>>  | f                | -1268658584
>>> 16400 | dn6       | D         |     15433 | xx.xx.40.123 | f
>>>  | f                | -1765597067
>>> 
>>> The entry with localhost was added after using the initdb command. I
>>> used it this way: initdb ­D CN1 - - node name CN1
>>> 
>>> The last troubleshooting step I tried as issuing and alter node to
>>> change the host of CN1 to it's IP instead of local host, but I get the
>>> ERROR:  Failed to get pooled connections error there too. I have tried
>>> to reload the pool several times with select pgxc_pool_reload(); That
>>> didn't work either.  I have added the entries in the pg_hba.conf for the
>>> two subnets and copied the hba file to all the servers and their nodes.
>>> 
>>> 
>>> Any help would be greatly appreciated. I'm certain it's something silly
>>> I did wrong in the cluster config, but am having trouble finding it.
>>> It's something very, very tiny I imagine. If there is any other config
>>> information needed let me know. Hope I've capture the most important
>>> bits.
>>> 
>>> Thanks!
>>> dew
>>> 
>>> 
>>> -------------------------------------------------------------------------
>>> -----
>>> HPCC Systems Open Source Big Data Platform from LexisNexis Risk
>>> Solutions
>>> Find What Matters Most in Your Big Data with HPCC Systems
>>> Open Source. Fast. Scalable. Simple. Ideal for Dirty Data.
>>> Leverages Graph Analysis for Fast Processing & Easy Data Exploration
>>> http://p.sf.net/sfu/hpccsystems
>>> _______________________________________________
>>> Postgres-xc-general mailing list
>>> Pos...@li...
>>> https://lists.sourceforge.net/lists/listinfo/postgres-xc-general
> 
> 
> ------------------------------------------------------------------------------
> HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions
> Find What Matters Most in Your Big Data with HPCC Systems
> Open Source. Fast. Scalable. Simple. Ideal for Dirty Data.
> Leverages Graph Analysis for Fast Processing & Easy Data Exploration
> http://p.sf.net/sfu/hpccsystems
> _______________________________________________
> Postgres-xc-general mailing list
> Pos...@li...
> https://lists.sourceforge.net/lists/listinfo/postgres-xc-general
> 

Thank you for the replies Masataka and Koichi. I appreciate them very much.

You were both right. It was definitely some config relates
issues/mismatches. I suppose my mind was pretty tired after configuring
each config file on everyone one of the servers and trying to troubleshoot
all day yesterday. Cross connecting with psql and using netstat definitely
helped suss out the problematic config items across the nodes and servers.

To fix my issue I made some corrections to pg_hba.conf for my
coordinators. My data nodes were configured correctly, but the
coordinators were not. I also noticed on CN1 and CN3, I missed changing
the listen address from localhost to * (0.0.0.0). That issue was apparent
with netstat -an

I'm going to have a look at the pgxc_ctl tool to simplify cluster starts
and stops as well as the config of future clusters for production use.
This is only a prototype environment.

Thanks again for the help my friends. I appreciate it.

dw

On 6/10/14 6:06 AM, "Masataka Saito" <pg...@gm...> wrote:

>Hello Wilkerson,
>
>I'm afraid that you're using Postgres-XC 1.2 which has such ugly bug
>fixed on 1.2.1.
>Or Postgres-XC nodes must trust other nodes: you have to add
>pg_hba.conf entry for subnet xx.xx.40.0/24 and xx.xx.33.0/24 with
>trust authentication method, so you can't specify other authentication
>method which requires extra dialogue with client, e.g. password or md5
>are not allowed.
>
>If both suggestions don't hit, please test that you can connect to
>other nodes from cn1 by psql command. I think it's a good way to find
>the component in which the cause resides.
>
>Regards.
>
>On 10 June 2014 07:13, Wilkerson, Daniel <dwi...@fu...> wrote:
>> Hi. I am trying to stand up a multi node cluster across multiple
>>servers. The current configuration I have is one coordinator and two
>>datanodes per server on 3 servers. Each has a GTM proxy configured. I
>>have two more servers configured to be the GTM master and slave/standby.
>>
>> When executing create database in my cluster after all the config work
>>(including the create node calls) I am getting the following error.
>>ERROR:  Failed to get pooled connections
>> CONTEXT:  SQL statement "EXECUTE DIRECT ON (cn2) 'SELECT
>>pg_catalog.pg_try_advisory_xact_lock_shared(65535, 0)'"
>>
>> I noticed that the coordinator on each node is added as host = local
>>host after using the initdb command to create the coordinator nodes.
>>When executing the create node calls to add the other coordinators and
>>datanodes to each other I am using the IP addresses of the servers
>>instead; I have also tried the DNS names also with no luck.
>>
>> The output of pgxc_node from one of the cluster coordinator/datanodes
>>looks like:
>>
>>  oid  | node_name | node_type | node_port |  node_host   |
>>nodeis_primary | nodeis_preferred |   node_id
>> 
>>-------+-----------+-----------+-----------+--------------+--------------
>>--+------------------+-------------
>>  11198 | cn1       | C         |      5432 | localhost    | f
>>   | f                | -1178713634
>>  16393 | cn2       | C         |      5432 | xx.xx.40.11  | f
>>   | f                | -1923125220
>>  16394 | cn3       | C         |      5432 | xx.xx.40.123 | f
>>   | f                |  1101067607
>>  16395 | dn1       | D         |     15432 | xx.xx.33.198 | t
>>   | t                |  -560021589
>>  16396 | dn2       | D         |     15433 | xx.xx.33.198 | f
>>   | f                |   352366662
>>  16397 | dn3       | D         |     15432 | xx.xx.40.11  | f
>>   | f                |  -700122826
>>  16398 | dn4       | D         |     15433 | xx.xx.40.11  | f
>>   | f                |   823103418
>>  16399 | dn5       | D         |     15432 | xx.xx.40.123 | f
>>   | f                | -1268658584
>>  16400 | dn6       | D         |     15433 | xx.xx.40.123 | f
>>   | f                | -1765597067
>>
>> The entry with localhost was added after using the initdb command. I
>>used it this way: initdb ­D CN1 - - node name CN1
>>
>> The last troubleshooting step I tried as issuing and alter node to
>>change the host of CN1 to it's IP instead of local host, but I get the
>>ERROR:  Failed to get pooled connections error there too. I have tried
>>to reload the pool several times with select pgxc_pool_reload(); That
>>didn't work either.  I have added the entries in the pg_hba.conf for the
>>two subnets and copied the hba file to all the servers and their nodes.
>>
>>
>> Any help would be greatly appreciated. I'm certain it's something silly
>>I did wrong in the cluster config, but am having trouble finding it.
>>It's something very, very tiny I imagine. If there is any other config
>>information needed let me know. Hope I've capture the most important
>>bits.
>>
>> Thanks!
>> dew
>>
>> 
>>-------------------------------------------------------------------------
>>-----
>> HPCC Systems Open Source Big Data Platform from LexisNexis Risk
>>Solutions
>> Find What Matters Most in Your Big Data with HPCC Systems
>> Open Source. Fast. Scalable. Simple. Ideal for Dirty Data.
>> Leverages Graph Analysis for Fast Processing & Easy Data Exploration
>> http://p.sf.net/sfu/hpccsystems
>> _______________________________________________
>> Postgres-xc-general mailing list
>> Pos...@li...
>> https://lists.sourceforge.net/lists/listinfo/postgres-xc-general

Hello Wilkerson,

I'm afraid that you're using Postgres-XC 1.2 which has such ugly bug
fixed on 1.2.1.
Or Postgres-XC nodes must trust other nodes: you have to add
pg_hba.conf entry for subnet xx.xx.40.0/24 and xx.xx.33.0/24 with
trust authentication method, so you can't specify other authentication
method which requires extra dialogue with client, e.g. password or md5
are not allowed.

If both suggestions don't hit, please test that you can connect to
other nodes from cn1 by psql command. I think it's a good way to find
the component in which the cause resides.

Regards.

On 10 June 2014 07:13, Wilkerson, Daniel <dwi...@fu...> wrote:
> Hi. I am trying to stand up a multi node cluster across multiple servers. The current configuration I have is one coordinator and two datanodes per server on 3 servers. Each has a GTM proxy configured. I have two more servers configured to be the GTM master and slave/standby.
>
> When executing create database in my cluster after all the config work (including the create node calls) I am getting the following error. ERROR:  Failed to get pooled connections
> CONTEXT:  SQL statement "EXECUTE DIRECT ON (cn2) 'SELECT pg_catalog.pg_try_advisory_xact_lock_shared(65535, 0)'"
>
> I noticed that the coordinator on each node is added as host = local host after using the initdb command to create the coordinator nodes. When executing the create node calls to add the other coordinators and datanodes to each other I am using the IP addresses of the servers instead; I have also tried the DNS names also with no luck.
>
> The output of pgxc_node from one of the cluster coordinator/datanodes looks like:
>
>  oid  | node_name | node_type | node_port |  node_host   | nodeis_primary | nodeis_preferred |   node_id
> -------+-----------+-----------+-----------+--------------+----------------+------------------+-------------
>  11198 | cn1       | C         |      5432 | localhost    | f              | f                | -1178713634
>  16393 | cn2       | C         |      5432 | xx.xx.40.11  | f              | f                | -1923125220
>  16394 | cn3       | C         |      5432 | xx.xx.40.123 | f              | f                |  1101067607
>  16395 | dn1       | D         |     15432 | xx.xx.33.198 | t              | t                |  -560021589
>  16396 | dn2       | D         |     15433 | xx.xx.33.198 | f              | f                |   352366662
>  16397 | dn3       | D         |     15432 | xx.xx.40.11  | f              | f                |  -700122826
>  16398 | dn4       | D         |     15433 | xx.xx.40.11  | f              | f                |   823103418
>  16399 | dn5       | D         |     15432 | xx.xx.40.123 | f              | f                | -1268658584
>  16400 | dn6       | D         |     15433 | xx.xx.40.123 | f              | f                | -1765597067
>
> The entry with localhost was added after using the initdb command. I used it this way: initdb –D CN1 - - node name CN1
>
> The last troubleshooting step I tried as issuing and alter node to change the host of CN1 to it's IP instead of local host, but I get the ERROR:  Failed to get pooled connections error there too. I have tried to reload the pool several times with select pgxc_pool_reload(); That didn't work either.  I have added the entries in the pg_hba.conf for the two subnets and copied the hba file to all the servers and their nodes.
>
>
> Any help would be greatly appreciated. I'm certain it's something silly I did wrong in the cluster config, but am having trouble finding it. It's something very, very tiny I imagine. If there is any other config information needed let me know. Hope I've capture the most important bits.
>
> Thanks!
> dew
>
> ------------------------------------------------------------------------------
> HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions
> Find What Matters Most in Your Big Data with HPCC Systems
> Open Source. Fast. Scalable. Simple. Ideal for Dirty Data.
> Leverages Graph Analysis for Fast Processing & Easy Data Exploration
> http://p.sf.net/sfu/hpccsystems
> _______________________________________________
> Postgres-xc-general mailing list
> Pos...@li...
> https://lists.sourceforge.net/lists/listinfo/postgres-xc-general

We had similar report in the past.   In this case, as you noticed, pg_hba.conf setting had some problem and setting up this correctly corrected the issue.    If you set log_min_messages to DEBUG1 or higher at all the nodes, you may get better hints.

BTY, you can correct pgxc_node settings with ALTER NODE.   Pgxc_ctl does this.   Also, please make sure that you ran {CREATE|ALTER} NODE at all the nodes.     Unlike other DDL, {CREATE | ALTER | DROP} NODE does not propagate to other nodes.

Regards;
---
Koichi Suzuki

2014/06/10 7:13、Wilkerson, Daniel <dwi...@fu...> のメール：

> Hi. I am trying to stand up a multi node cluster across multiple servers. The current configuration I have is one coordinator and two datanodes per server on 3 servers. Each has a GTM proxy configured. I have two more servers configured to be the GTM master and slave/standby.
> 
> When executing create database in my cluster after all the config work (including the create node calls) I am getting the following error. ERROR:  Failed to get pooled connections
> CONTEXT:  SQL statement "EXECUTE DIRECT ON (cn2) 'SELECT pg_catalog.pg_try_advisory_xact_lock_shared(65535, 0)'"
> 
> I noticed that the coordinator on each node is added as host = local host after using the initdb command to create the coordinator nodes. When executing the create node calls to add the other coordinators and datanodes to each other I am using the IP addresses of the servers instead; I have also tried the DNS names also with no luck.
> 
> The output of pgxc_node from one of the cluster coordinator/datanodes looks like:
> 
> oid  | node_name | node_type | node_port |  node_host   | nodeis_primary | nodeis_preferred |   node_id
> -------+-----------+-----------+-----------+--------------+----------------+------------------+-------------
> 11198 | cn1       | C         |      5432 | localhost    | f              | f                | -1178713634
> 16393 | cn2       | C         |      5432 | xx.xx.40.11  | f              | f                | -1923125220
> 16394 | cn3       | C         |      5432 | xx.xx.40.123 | f              | f                |  1101067607
> 16395 | dn1       | D         |     15432 | xx.xx.33.198 | t              | t                |  -560021589
> 16396 | dn2       | D         |     15433 | xx.xx.33.198 | f              | f                |   352366662
> 16397 | dn3       | D         |     15432 | xx.xx.40.11  | f              | f                |  -700122826
> 16398 | dn4       | D         |     15433 | xx.xx.40.11  | f              | f                |   823103418
> 16399 | dn5       | D         |     15432 | xx.xx.40.123 | f              | f                | -1268658584
> 16400 | dn6       | D         |     15433 | xx.xx.40.123 | f              | f                | -1765597067
> 
> The entry with localhost was added after using the initdb command. I used it this way: initdb –D CN1 - - node name CN1
> 
> The last troubleshooting step I tried as issuing and alter node to change the host of CN1 to it's IP instead of local host, but I get the ERROR:  Failed to get pooled connections error there too. I have tried to reload the pool several times with select pgxc_pool_reload(); That didn't work either.  I have added the entries in the pg_hba.conf for the two subnets and copied the hba file to all the servers and their nodes.
> 
> 
> Any help would be greatly appreciated. I'm certain it's something silly I did wrong in the cluster config, but am having trouble finding it. It's something very, very tiny I imagine. If there is any other config information needed let me know. Hope I've capture the most important bits.
> 
> Thanks!
> dew
> 
> ------------------------------------------------------------------------------
> HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions
> Find What Matters Most in Your Big Data with HPCC Systems
> Open Source. Fast. Scalable. Simple. Ideal for Dirty Data.
> Leverages Graph Analysis for Fast Processing & Easy Data Exploration
> http://p.sf.net/sfu/hpccsystems
> _______________________________________________
> Postgres-xc-general mailing list
> Pos...@li...
> https://lists.sourceforge.net/lists/listinfo/postgres-xc-general
> 

Hi. I am trying to stand up a multi node cluster across multiple servers. The current configuration I have is one coordinator and two datanodes per server on 3 servers. Each has a GTM proxy configured. I have two more servers configured to be the GTM master and slave/standby.

When executing create database in my cluster after all the config work (including the create node calls) I am getting the following error. ERROR:  Failed to get pooled connections
CONTEXT:  SQL statement "EXECUTE DIRECT ON (cn2) 'SELECT pg_catalog.pg_try_advisory_xact_lock_shared(65535, 0)'"

I noticed that the coordinator on each node is added as host = local host after using the initdb command to create the coordinator nodes. When executing the create node calls to add the other coordinators and datanodes to each other I am using the IP addresses of the servers instead; I have also tried the DNS names also with no luck.

The output of pgxc_node from one of the cluster coordinator/datanodes looks like:

 oid  | node_name | node_type | node_port |  node_host   | nodeis_primary | nodeis_preferred |   node_id
-------+-----------+-----------+-----------+--------------+----------------+------------------+-------------
 11198 | cn1       | C         |      5432 | localhost    | f              | f                | -1178713634
 16393 | cn2       | C         |      5432 | xx.xx.40.11  | f              | f                | -1923125220
 16394 | cn3       | C         |      5432 | xx.xx.40.123 | f              | f                |  1101067607
 16395 | dn1       | D         |     15432 | xx.xx.33.198 | t              | t                |  -560021589
 16396 | dn2       | D         |     15433 | xx.xx.33.198 | f              | f                |   352366662
 16397 | dn3       | D         |     15432 | xx.xx.40.11  | f              | f                |  -700122826
 16398 | dn4       | D         |     15433 | xx.xx.40.11  | f              | f                |   823103418
 16399 | dn5       | D         |     15432 | xx.xx.40.123 | f              | f                | -1268658584
 16400 | dn6       | D         |     15433 | xx.xx.40.123 | f              | f                | -1765597067

The entry with localhost was added after using the initdb command. I used it this way: initdb –D CN1 - - node name CN1

The last troubleshooting step I tried as issuing and alter node to change the host of CN1 to it's IP instead of local host, but I get the ERROR:  Failed to get pooled connections error there too. I have tried to reload the pool several times with select pgxc_pool_reload(); That didn't work either.  I have added the entries in the pg_hba.conf for the two subnets and copied the hba file to all the servers and their nodes.


Any help would be greatly appreciated. I'm certain it's something silly I did wrong in the cluster config, but am having trouble finding it. It's something very, very tiny I imagine. If there is any other config information needed let me know. Hope I've capture the most important bits.

Thanks!
dew

More robust node redundancy and read-only slave will be one of the next issue.

Hope to have much more input on this.    I’m going to open next roadmap discussion.
---
Koichi Suzuki

2014/06/03 3:33、Mason Sharp <ms...@tr...<mailto:ms...@tr...>> のメール：




On Sun, May 25, 2014 at 2:04 PM, Koichi Suzuki <koi...@gm...<mailto:koi...@gm...>> wrote:
I see.  Your have good usecase for read-only transactions.

Because of the nature of log shipping and sharing/clustering, it is
not simple to provide read-only transaction in XC.

There may be cases where it is being used as a data warehouse (thinking a bit more about Postgres-XL here), where there is a nightly load process and then reads are done during the day.  If the standbys are dedicated, being able to load balance them could be useful.


 Two essential
reasons:

1. Delay in WAL playback in each slave may be different.   It makes
providing consistent database view extremely difficult.


2. At present, slave calculates snapshot of the transaction from the
WAL.   Current code does not allow missing XIDs.   There will be
memory leak and crash by OOM if there's many missing XIDs in the WAL
stream.  In XC, it is disabled and the database view may be
inconsistent.   Please note that this does not affect recovery and
promotion.



Agreed, there may be some level of effort for this.

--
Mason Sharp

TransLattice - http://www.translattice.com<http://www.translattice.com/>
Distributed and Clustered Database Solutions


------------------------------------------------------------------------------
Learn Graph Databases - Download FREE O'Reilly Book
"Graph Databases" is the definitive new guide to graph databases and their
applications. Written by three acclaimed leaders in the field,
this first edition is now available. Download your free book today!
http://p.sf.net/sfu/NeoTech_______________________________________________
Postgres-xc-general mailing list
Pos...@li...
https://lists.sourceforge.net/lists/listinfo/postgres-xc-general

Hi,

I reviewed the patch.

+1 to go to master and 1.x.

Regards;
---
Koichi Suzuki

2014/06/03 17:31、Masataka Saito <pg...@gm...> のメール：

> Hi, Aaron
> 
> I think you've done right analysis.
> 
> The order of evaluation of sub-expressions and the order in which side
> effects take place are frequently defined as unspecified behavior by
> the C Standard. As you saw, it's one of the such case that the order
> in which the arguments to a function are evaluated.
> 
> The patch looks good, and I think we must back-patch it to all the
> versions after 1.0.
> 
> Regards.
> 
> On 3 June 2014 16:20, Aaron Jackson <aja...@re...> wrote:
>> I've been able to work my way backwards through the problem and have
>> discovered the underlying problem.  When a data coordinator is paired with a
>> GTM proxy, it forwards its message to the GTM proxy who adds some data to
>> the payload and forwards it to the GTM.  Here is what I saw when looking at
>> the wire.
>> 
>> The message captured between the data coordinator and the GTM proxy was as
>> follows:
>> 
>> 430000003d000000270000001064656d6f2e7075626c69632e666f6f0001000000000000000100000000000000ffffffffffffff7f010000000000000000
>> 
>> The message captured between the GTM proxy and the GTM was as follows:
>> 
>> 430000000a00000000002746000000080000003b
>> 
>> 
>> Definitely a horrible truncation of the payload.  The problem is in
>> GTMProxy_ProxyCommand, specifically, the two calls to pq_getmsgunreadlen().
>> The assumption is that these are called before anything else.
>> Unfortunately, the intel compiler calls pq_getmsgbytes() and subsequently
>> calls the second instance of pq_getmsgunreadlen().  The second time it is
>> called, the value returns zero and we end up with all kinds of byte
>> truncation.  I've attached a patch to fix the issue.
>> 
>> --- postgres-xc-1.2.1-orig/src/gtm/proxy/proxy_main.c   2014-04-03
>> 05:18:38.000000000 +0000
>> +++ postgres-xc-1.2.1/src/gtm/proxy/proxy_main.c        2014-06-03
>> 07:14:58.451411000 +0000
>> @@ -2390,6 +2390,7 @@
>>        GTMProxy_CommandInfo *cmdinfo;
>>        GTMProxy_ThreadInfo *thrinfo = GetMyThreadInfo;
>>        GTM_ProxyMsgHeader proxyhdr;
>> +       size_t msgunreadlen = pq_getmsgunreadlen(message);
>> 
>>        proxyhdr.ph_conid = conninfo->con_id;
>> 
>> @@ -2397,8 +2398,8 @@
>>        if (gtmpqPutMsgStart('C', true, gtm_conn) ||
>>                gtmpqPutnchar((char *)&proxyhdr, sizeof
>> (GTM_ProxyMsgHeader), gtm_conn) ||
>>                gtmpqPutInt(mtype, sizeof (GTM_MessageType), gtm_conn) ||
>> -               gtmpqPutnchar(pq_getmsgbytes(message,
>> pq_getmsgunreadlen(message)),
>> -                                         pq_getmsgunreadlen(message),
>> gtm_conn))
>> +               gtmpqPutnchar(pq_getmsgbytes(message, msgunreadlen),
>> +                                         msgunreadlen, gtm_conn))
>>                elog(ERROR, "Error proxing data");
>> 
>>        /*
>> 
>> 
>> 
>> Aaron
>> ________________________________
>> From: Aaron Jackson [aja...@re...]
>> Sent: Monday, June 02, 2014 4:11 PM
>> To: pos...@li...
>> Subject: [Postgres-xc-general] Unable to create sequences
>> 
>> I tried to create a database as follows ...
>> 
>> CREATE TABLE Schema.TableFoo(
>>    SomeId serial NOT NULL,
>>    ForeignId int NOT NULL,
>>    ...
>> ) WITH (OIDS = FALSE);
>> 
>> 
>> The server returned the following...
>> 
>> ERROR:  GTM error, could not create sequence
>> 
>> 
>> Looked at the server logs for the gtm_proxy, nothing so I went to the gtm.
>> 
>> LOCATION:  pq_copymsgbytes, pqformat.c:554
>> 1:140488486782720:2014-06-02 21:10:58.870 UTC -WARNING:  No transaction
>> handle for gxid: 0
>> LOCATION:  GTM_GXIDToHandle, gtm_txn.c:163
>> 1:140488486782720:2014-06-02 21:10:58.870 UTC -WARNING:  Invalid transaction
>> handle: -1
>> LOCATION:  GTM_HandleToTransactionInfo, gtm_txn.c:213
>> 1:140488486782720:2014-06-02 21:10:58.870 UTC -ERROR:  Failed to get a
>> snapshot
>> LOCATION:  ProcessGetSnapshotCommandMulti, gtm_snap.c:420
>> 1:140488478390016:2014-06-02 21:10:58.871 UTC -ERROR:  insufficient data
>> left in message
>> LOCATION:  pq_copymsgbytes, pqformat.c:554
>> 1:140488486782720:2014-06-02 21:10:58.871 UTC -ERROR:  insufficient data
>> left in message
>> LOCATION:  pq_copymsgbytes, pqformat.c:554
>> 
>> 
>> I'm definitely confused here.  This cluster has been running fine for
>> several days now.  And now the GTM is failing. I performed a restart of the
>> gtm and proxies (each using gtm_ctl to stop and restart the instance).
>> Nothing has changed, the GTM continues to fail and will not create the
>> sequence.
>> 
>> Any ideas?
>> 
>> Aaron
>> 
>> ------------------------------------------------------------------------------
>> Learn Graph Databases - Download FREE O'Reilly Book
>> "Graph Databases" is the definitive new guide to graph databases and their
>> applications. Written by three acclaimed leaders in the field,
>> this first edition is now available. Download your free book today!
>> http://p.sf.net/sfu/NeoTech
>> _______________________________________________
>> Postgres-xc-general mailing list
>> Pos...@li...
>> https://lists.sourceforge.net/lists/listinfo/postgres-xc-general
>> 
> 
> ------------------------------------------------------------------------------
> Learn Graph Databases - Download FREE O'Reilly Book
> "Graph Databases" is the definitive new guide to graph databases and their 
> applications. Written by three acclaimed leaders in the field, 
> this first edition is now available. Download your free book today!
> http://p.sf.net/sfu/NeoTech
> _______________________________________________
> Postgres-xc-general mailing list
> Pos...@li...
> https://lists.sourceforge.net/lists/listinfo/postgres-xc-general
> 

Ok!

Thank you!



2014-06-04 0:53 GMT-03:00 Michael Paquier <mic...@gm...>:

> On Wed, Jun 4, 2014 at 3:50 AM, Bruno Cezar Oliveira
> <bru...@ho...> wrote:
> > How i can pooling the connections on the coordinators using java? PGpool?
> With java? Postgres JDBC as it has support for multiple sources:
> jdbc:postgresql://host1:port1,host2:port2/test
> Support is basic and connections are tried in a round robin fashion
> until one connection is successful.
> --
> Michael
>
>


-- 
Att,
Bruno Cezar de Oliveira.
bru...@ho...
-------------------------------------------------------
Uberlândia - MG.

On Wed, Jun 4, 2014 at 3:50 AM, Bruno Cezar Oliveira
<bru...@ho...> wrote:
> How i can pooling the connections on the coordinators using java? PGpool?
With java? Postgres JDBC as it has support for multiple sources:
jdbc:postgresql://host1:port1,host2:port2/test
Support is basic and connections are tried in a round robin fashion
until one connection is successful.
-- 
Michael

Hi,

How i can pooling the connections on the coordinators using java? PGpool?


Thanks!

-- 
Att,
Bruno Cezar de Oliveira.
bru...@ho...
-------------------------------------------------------
Uberlândia - MG.

Hi, Aaron

I think you've done right analysis.

The order of evaluation of sub-expressions and the order in which side
effects take place are frequently defined as unspecified behavior by
the C Standard. As you saw, it's one of the such case that the order
in which the arguments to a function are evaluated.

The patch looks good, and I think we must back-patch it to all the
versions after 1.0.

Regards.

On 3 June 2014 16:20, Aaron Jackson <aja...@re...> wrote:
> I've been able to work my way backwards through the problem and have
> discovered the underlying problem.  When a data coordinator is paired with a
> GTM proxy, it forwards its message to the GTM proxy who adds some data to
> the payload and forwards it to the GTM.  Here is what I saw when looking at
> the wire.
>
> The message captured between the data coordinator and the GTM proxy was as
> follows:
>
> 430000003d000000270000001064656d6f2e7075626c69632e666f6f0001000000000000000100000000000000ffffffffffffff7f010000000000000000
>
> The message captured between the GTM proxy and the GTM was as follows:
>
> 430000000a00000000002746000000080000003b
>
>
> Definitely a horrible truncation of the payload.  The problem is in
> GTMProxy_ProxyCommand, specifically, the two calls to pq_getmsgunreadlen().
> The assumption is that these are called before anything else.
> Unfortunately, the intel compiler calls pq_getmsgbytes() and subsequently
> calls the second instance of pq_getmsgunreadlen().  The second time it is
> called, the value returns zero and we end up with all kinds of byte
> truncation.  I've attached a patch to fix the issue.
>
> --- postgres-xc-1.2.1-orig/src/gtm/proxy/proxy_main.c   2014-04-03
> 05:18:38.000000000 +0000
> +++ postgres-xc-1.2.1/src/gtm/proxy/proxy_main.c        2014-06-03
> 07:14:58.451411000 +0000
> @@ -2390,6 +2390,7 @@
>         GTMProxy_CommandInfo *cmdinfo;
>         GTMProxy_ThreadInfo *thrinfo = GetMyThreadInfo;
>         GTM_ProxyMsgHeader proxyhdr;
> +       size_t msgunreadlen = pq_getmsgunreadlen(message);
>
>         proxyhdr.ph_conid = conninfo->con_id;
>
> @@ -2397,8 +2398,8 @@
>         if (gtmpqPutMsgStart('C', true, gtm_conn) ||
>                 gtmpqPutnchar((char *)&proxyhdr, sizeof
> (GTM_ProxyMsgHeader), gtm_conn) ||
>                 gtmpqPutInt(mtype, sizeof (GTM_MessageType), gtm_conn) ||
> -               gtmpqPutnchar(pq_getmsgbytes(message,
> pq_getmsgunreadlen(message)),
> -                                         pq_getmsgunreadlen(message),
> gtm_conn))
> +               gtmpqPutnchar(pq_getmsgbytes(message, msgunreadlen),
> +                                         msgunreadlen, gtm_conn))
>                 elog(ERROR, "Error proxing data");
>
>         /*
>
>
>
> Aaron
> ________________________________
> From: Aaron Jackson [aja...@re...]
> Sent: Monday, June 02, 2014 4:11 PM
> To: pos...@li...
> Subject: [Postgres-xc-general] Unable to create sequences
>
> I tried to create a database as follows ...
>
> CREATE TABLE Schema.TableFoo(
>     SomeId serial NOT NULL,
>     ForeignId int NOT NULL,
>     ...
> ) WITH (OIDS = FALSE);
>
>
> The server returned the following...
>
> ERROR:  GTM error, could not create sequence
>
>
> Looked at the server logs for the gtm_proxy, nothing so I went to the gtm.
>
> LOCATION:  pq_copymsgbytes, pqformat.c:554
> 1:140488486782720:2014-06-02 21:10:58.870 UTC -WARNING:  No transaction
> handle for gxid: 0
> LOCATION:  GTM_GXIDToHandle, gtm_txn.c:163
> 1:140488486782720:2014-06-02 21:10:58.870 UTC -WARNING:  Invalid transaction
> handle: -1
> LOCATION:  GTM_HandleToTransactionInfo, gtm_txn.c:213
> 1:140488486782720:2014-06-02 21:10:58.870 UTC -ERROR:  Failed to get a
> snapshot
> LOCATION:  ProcessGetSnapshotCommandMulti, gtm_snap.c:420
> 1:140488478390016:2014-06-02 21:10:58.871 UTC -ERROR:  insufficient data
> left in message
> LOCATION:  pq_copymsgbytes, pqformat.c:554
> 1:140488486782720:2014-06-02 21:10:58.871 UTC -ERROR:  insufficient data
> left in message
> LOCATION:  pq_copymsgbytes, pqformat.c:554
>
>
> I'm definitely confused here.  This cluster has been running fine for
> several days now.  And now the GTM is failing. I performed a restart of the
> gtm and proxies (each using gtm_ctl to stop and restart the instance).
> Nothing has changed, the GTM continues to fail and will not create the
> sequence.
>
> Any ideas?
>
> Aaron
>
> ------------------------------------------------------------------------------
> Learn Graph Databases - Download FREE O'Reilly Book
> "Graph Databases" is the definitive new guide to graph databases and their
> applications. Written by three acclaimed leaders in the field,
> this first edition is now available. Download your free book today!
> http://p.sf.net/sfu/NeoTech
> _______________________________________________
> Postgres-xc-general mailing list
> Pos...@li...
> https://lists.sourceforge.net/lists/listinfo/postgres-xc-general
>

I've been able to work my way backwards through the problem and have discovered the underlying problem.  When a data coordinator is paired with a GTM proxy, it forwards its message to the GTM proxy who adds some data to the payload and forwards it to the GTM.  Here is what I saw when looking at the wire.

The message captured between the data coordinator and the GTM proxy was as follows:
430000003d000000270000001064656d6f2e7075626c69632e666f6f0001000000000000000100000000000000ffffffffffffff7f010000000000000000
The message captured between the GTM proxy and the GTM was as follows:
430000000a00000000002746000000080000003b

Definitely a horrible truncation of the payload.  The problem is in GTMProxy_ProxyCommand, specifically, the two calls to pq_getmsgunreadlen().  The assumption is that these are called before anything else.  Unfortunately, the intel compiler calls pq_getmsgbytes() and subsequently calls the second instance of pq_getmsgunreadlen().  The second time it is called, the value returns zero and we end up with all kinds of byte truncation.  I've attached a patch to fix the issue.

--- postgres-xc-1.2.1-orig/src/gtm/proxy/proxy_main.c   2014-04-03 05:18:38.000000000 +0000
+++ postgres-xc-1.2.1/src/gtm/proxy/proxy_main.c        2014-06-03 07:14:58.451411000 +0000
@@ -2390,6 +2390,7 @@
        GTMProxy_CommandInfo *cmdinfo;
        GTMProxy_ThreadInfo *thrinfo = GetMyThreadInfo;
        GTM_ProxyMsgHeader proxyhdr;
+       size_t msgunreadlen = pq_getmsgunreadlen(message);

        proxyhdr.ph_conid = conninfo->con_id;

@@ -2397,8 +2398,8 @@
        if (gtmpqPutMsgStart('C', true, gtm_conn) ||
                gtmpqPutnchar((char *)&proxyhdr, sizeof (GTM_ProxyMsgHeader), gtm_conn) ||
                gtmpqPutInt(mtype, sizeof (GTM_MessageType), gtm_conn) ||
-               gtmpqPutnchar(pq_getmsgbytes(message, pq_getmsgunreadlen(message)),
-                                         pq_getmsgunreadlen(message), gtm_conn))
+               gtmpqPutnchar(pq_getmsgbytes(message, msgunreadlen),
+                                         msgunreadlen, gtm_conn))
                elog(ERROR, "Error proxing data");

        /*


Aaron
________________________________
From: Aaron Jackson [aja...@re...]
Sent: Monday, June 02, 2014 4:11 PM
To: pos...@li...
Subject: [Postgres-xc-general] Unable to create sequences

I tried to create a database as follows ...

CREATE TABLE Schema.TableFoo(
    SomeId serial NOT NULL,
    ForeignId int NOT NULL,
    ...
) WITH (OIDS = FALSE);

The server returned the following...

ERROR:  GTM error, could not create sequence

Looked at the server logs for the gtm_proxy, nothing so I went to the gtm.

LOCATION:  pq_copymsgbytes, pqformat.c:554
1:140488486782720:2014-06-02 21:10:58.870 UTC -WARNING:  No transaction handle for gxid: 0
LOCATION:  GTM_GXIDToHandle, gtm_txn.c:163
1:140488486782720:2014-06-02 21:10:58.870 UTC -WARNING:  Invalid transaction handle: -1
LOCATION:  GTM_HandleToTransactionInfo, gtm_txn.c:213
1:140488486782720:2014-06-02 21:10:58.870 UTC -ERROR:  Failed to get a snapshot
LOCATION:  ProcessGetSnapshotCommandMulti, gtm_snap.c:420
1:140488478390016:2014-06-02 21:10:58.871 UTC -ERROR:  insufficient data left in message
LOCATION:  pq_copymsgbytes, pqformat.c:554
1:140488486782720:2014-06-02 21:10:58.871 UTC -ERROR:  insufficient data left in message
LOCATION:  pq_copymsgbytes, pqformat.c:554

I'm definitely confused here.  This cluster has been running fine for several days now.  And now the GTM is failing. I performed a restart of the gtm and proxies (each using gtm_ctl to stop and restart the instance).  Nothing has changed, the GTM continues to fail and will not create the sequence.

Any ideas?

Aaron

I tried to create a database as follows ...

CREATE TABLE Schema.TableFoo(
    SomeId serial NOT NULL,
    ForeignId int NOT NULL,
    ...
) WITH (OIDS = FALSE);

The server returned the following...

ERROR:  GTM error, could not create sequence

Looked at the server logs for the gtm_proxy, nothing so I went to the gtm.

LOCATION:  pq_copymsgbytes, pqformat.c:554
1:140488486782720:2014-06-02 21:10:58.870 UTC -WARNING:  No transaction handle for gxid: 0
LOCATION:  GTM_GXIDToHandle, gtm_txn.c:163
1:140488486782720:2014-06-02 21:10:58.870 UTC -WARNING:  Invalid transaction handle: -1
LOCATION:  GTM_HandleToTransactionInfo, gtm_txn.c:213
1:140488486782720:2014-06-02 21:10:58.870 UTC -ERROR:  Failed to get a snapshot
LOCATION:  ProcessGetSnapshotCommandMulti, gtm_snap.c:420
1:140488478390016:2014-06-02 21:10:58.871 UTC -ERROR:  insufficient data left in message
LOCATION:  pq_copymsgbytes, pqformat.c:554
1:140488486782720:2014-06-02 21:10:58.871 UTC -ERROR:  insufficient data left in message
LOCATION:  pq_copymsgbytes, pqformat.c:554

I'm definitely confused here.  This cluster has been running fine for several days now.  And now the GTM is failing. I performed a restart of the gtm and proxies (each using gtm_ctl to stop and restart the instance).  Nothing has changed, the GTM continues to fail and will not create the sequence.

Any ideas?

Aaron

On Sun, May 25, 2014 at 2:04 PM, Koichi Suzuki <koi...@gm...>
wrote:

> I see.  Your have good usecase for read-only transactions.
>
> Because of the nature of log shipping and sharing/clustering, it is
> not simple to provide read-only transaction in XC.


There may be cases where it is being used as a data warehouse (thinking a
bit more about Postgres-XL here), where there is a nightly load process and
then reads are done during the day.  If the standbys are dedicated, being
able to load balance them could be useful.



>  Two essential
> reasons:
>
> 1. Delay in WAL playback in each slave may be different.   It makes
> providing consistent database view extremely difficult.
>


> 2. At present, slave calculates snapshot of the transaction from the
> WAL.   Current code does not allow missing XIDs.   There will be
> memory leak and crash by OOM if there's many missing XIDs in the WAL
> stream.  In XC, it is disabled and the database view may be
> inconsistent.   Please note that this does not affect recovery and
> promotion.
>
>
>
Agreed, there may be some level of effort for this.

-- 
Mason Sharp

TransLattice - http://www.translattice.com
Distributed and Clustered Database Solutions

On Thu, May 29, 2014 at 7:27 AM, Alan Bryant <ala...@hd...> wrote:

>
>
>  - Can we shard on schema? (i.e. put some schemas in one location and
> other schemas in different locations, sharded the same way rows are?) Or if
> not, can we shard one table, and have functions execute on the same node as
> some controlling row?
>
>
>  No.
>
>    Interesting. This leaves us with the "one big table" approach, just
> spread across the data nodes. Aren't there performance issues with having
> massive individual tables? What techniques exist to help with this?
>

The individual tables can be replicated or sharded, such as a hash based on
one of the columns in the table.

A parent-child type of relation can be sharded on a primary key of the
parent, and the foreign key referencing that table in the child. The
planner recognizes that this data will co-exist on the same node and
"push-down" joins.

Other tables can be replicated, with an exact copy on each node if desired.
Here joins can always be pushed down. If a query only involves replicated
tables, it will just be sent to one single node. If a coordinator and data
node are on the same server, one can configure the system to prefer to get
this data from a local data node.


>
> Partitioning is one option I guess, using a different (but compatible)
> scheme than the sharding key, perhaps a different bitmask on the same key.
>

At the moment users cannot specify an arbitrary expression like a bitmask.
A modulo option exists, however. I think this is an item that should be a
relatively high-ish priority on the to do list.


>
> Another question is upgrade. We use schemas primarily to make logical
> upgrades possible while online, but what about Postgres-XC itself... what
> is the upgrade model for Postgres-XC? (say, between major PostgreSQL
> versions)
>
>
This is another area that can be improved, leveraging pg_upgrade. For now,
dump and restore...



> Thanks!
> -Alan
>

Regards,

-- 
Mason Sharp

TransLattice - http://www.translattice.com
Distributed and Clustered Database Solutions