postgres-xc-committers Mailing List for Postgres-XC

Project "Postgres-XC".

The branch, master has been updated
       via  fe9985c168d85738e5d88ed9407b840449f31b75 (commit)
      from  ecfaffc851a29902a5169c44358d5290cea0c627 (commit)


- Log -----------------------------------------------------------------
http://postgres-xc.git.sourceforge.net/git/gitweb.cgi?p=postgres-xc/postgres-xc;a=commitdiff;h=fe9985c168d85738e5d88ed9407b840449f31b75

commit fe9985c168d85738e5d88ed9407b840449f31b75
Author: Michael Paquier <mi...@ot...>
Date:   Tue Aug 14 15:59:05 2012 +0900

    Fix snapshot management on slave nodes and Datanodes accessed directly by applications
    
    The following things are changed:
    - Make a slave node session acquire its snapshot directly from WAL
    like PostgreSQL does. We do not try to get a global snapshot from GTM
    - Allow a Datanode to get a snapshot directly from GTM when accessed directly
    by an external application
    - Block run of queries interacting with remote nodes on a slave nodes.
    In this case it is necessary to obtain a transaction ID but a node in
    recovery cannot do that.
    
    in consequence, the following issues are fixed:
    - Allow read queries to run on Coordinator slaves by allowing getting a
    snaphot. A slave Coordinator was complaining about the impossibility to get
    a transaction ID when fetching a snapshot for all types of queries. This allows
    to check if the slave Coordinator is still alive and to monitor its statistics
    as long as the query does not involve data located on remote Datanodes.
    - Suppression of the annoying WARNING "Cannot get a global snapshot from GTM"
    when connecting to a Datanode directly with a psql client. Now in this case
    a snapshot is taken directly from GTM. Processes like autovacuum did that already.
    - Make slave nodes getting there snapshots directly from WAL and not from GTM.
    
    This is also backpatched to 1.0.

M	src/backend/access/transam/varsup.c
M	src/backend/access/transam/xact.c
M	src/backend/pgxc/pool/execRemote.c
M	src/backend/storage/ipc/procarray.c
M	src/include/access/xact.h

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

Summary of changes:
 src/backend/access/transam/varsup.c |   38 ++++++------
 src/backend/access/transam/xact.c   |   20 +++++-
 src/backend/pgxc/pool/execRemote.c  |    7 ++
 src/backend/storage/ipc/procarray.c |  120 ++++++++++++++++++++++-------------
 src/include/access/xact.h           |    1 +
 5 files changed, 121 insertions(+), 65 deletions(-)


hooks/post-receive
-- 
Postgres-XC

Project "Postgres-XC".

The branch, REL1_0_STABLE has been updated
       via  0449947063316f6108bddae1fda76006f85d6317 (commit)
      from  de6bebfed94ed6022181e4e11ac3c72345f81b8c (commit)


- Log -----------------------------------------------------------------
http://postgres-xc.git.sourceforge.net/git/gitweb.cgi?p=postgres-xc/postgres-xc;a=commitdiff;h=0449947063316f6108bddae1fda76006f85d6317

commit fe9985c168d85738e5d88ed9407b840449f31b75
Author: Michael Paquier <mi...@ot...>
Date:   Tue Aug 14 15:59:05 2012 +0900

    Fix snapshot management on slave nodes and Datanodes accessed directly by applications
    
    The following things are changed:
    - Make a slave node session acquire its snapshot directly from WAL
    like PostgreSQL does. We do not try to get a global snapshot from GTM
    - Allow a Datanode to get a snapshot directly from GTM when accessed directly
    by an external application
    - Block run of queries interacting with remote nodes on a slave nodes.
    In this case it is necessary to obtain a transaction ID but a node in
    recovery cannot do that.
    
    in consequence, the following issues are fixed:
    - Allow read queries to run on Coordinator slaves by allowing getting a
    snaphot. A slave Coordinator was complaining about the impossibility to get
    a transaction ID when fetching a snapshot for all types of queries. This allows
    to check if the slave Coordinator is still alive and to monitor its statistics
    as long as the query does not involve data located on remote Datanodes.
    - Suppression of the annoying WARNING "Cannot get a global snapshot from GTM"
    when connecting to a Datanode directly with a psql client. Now in this case
    a snapshot is taken directly from GTM. Processes like autovacuum did that already.
    - Make slave nodes getting there snapshots directly from WAL and not from GTM.
    
    This is also backpatched to 1.0.

M	src/backend/access/transam/varsup.c
M	src/backend/access/transam/xact.c
M	src/backend/pgxc/pool/execRemote.c
M	src/backend/storage/ipc/procarray.c
M	src/include/access/xact.h

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

Summary of changes:
 src/backend/access/transam/varsup.c |   38 ++++----
 src/backend/access/transam/xact.c   |   53 +++++++++----
 src/backend/pgxc/pool/execRemote.c  |    7 ++
 src/backend/storage/ipc/procarray.c |  156 +++++++++++++++++++++--------------
 src/include/access/xact.h           |    1 +
 5 files changed, 160 insertions(+), 95 deletions(-)


hooks/post-receive
-- 
Postgres-XC

Project "Postgres-XC".

The branch, master has been updated
       via  ecfaffc851a29902a5169c44358d5290cea0c627 (commit)
      from  cf649481d5d04cb89b1c3632cf998a63d6ebfa74 (commit)


- Log -----------------------------------------------------------------
http://postgres-xc.git.sourceforge.net/git/gitweb.cgi?p=postgres-xc/postgres-xc;a=commitdiff;h=ecfaffc851a29902a5169c44358d5290cea0c627

commit ecfaffc851a29902a5169c44358d5290cea0c627
Author: Michael <mi...@ot...>
Date:   Tue Aug 14 15:27:41 2012 +0900

    Add a forgotten \n in --help of pgxc_monitor
    
    This made --help call more difficult to read.

M	contrib/pgxc_monitor/pgxc_monitor.c

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

Summary of changes:
 contrib/pgxc_monitor/pgxc_monitor.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)


hooks/post-receive
-- 
Postgres-XC

Project "Postgres-XC".

The branch, master has been updated
       via  cf649481d5d04cb89b1c3632cf998a63d6ebfa74 (commit)
      from  f398526aaad7fe681c37f1fd2bda1d2953bc3c27 (commit)


- Log -----------------------------------------------------------------
http://postgres-xc.git.sourceforge.net/git/gitweb.cgi?p=postgres-xc/postgres-xc;a=commitdiff;h=cf649481d5d04cb89b1c3632cf998a63d6ebfa74

commit cf649481d5d04cb89b1c3632cf998a63d6ebfa74
Author: Michael Paquier <mi...@ot...>
Date:   Tue Aug 14 15:24:54 2012 +0900

    Refactor pgxc_monitor: help option bug and overall review
    
    pgxc_monitor was not printing help even if requested by "--help"
    or "-?". This has been reported by Nikhil Sontakke but I did not
    use his patch.
    Some whitespaces and spelling mistakes in code are corrected.
    Documentation had also some mistakes but it is fixed now.
    The creation of the psql command used to monitor nodes is made
    more readable and now facilitates future extensions.

M	contrib/pgxc_monitor/pgxc_monitor.c
M	doc-xc/src/sgml/pgxcmonitor.sgmlin

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

Summary of changes:
 contrib/pgxc_monitor/pgxc_monitor.c |  142 +++++++++++++++++++++--------------
 doc-xc/src/sgml/pgxcmonitor.sgmlin  |   20 +++---
 2 files changed, 95 insertions(+), 67 deletions(-)


hooks/post-receive
-- 
Postgres-XC