[pgpool-general: 8779] Re: Difference between pgpool 'primary' and Watchdog 'MASTER' versus the 'load_balance_node'

Tatsuo Ishii ishii at sraoss.co.jp
Mon May 29 17:37:31 JST 2023


> Hi pgpool experts,
> 
> Can you please clarify the below?
> *During failover, Is the '*primary*' or '*MASTER*' or ' *load_balance_node*
> failure taken into account?*

It's hard to answer your question as the question is too vague. Anyway I try it...

- primary failure is taken into account because it triggers failover.
  See https://www.pgpool.net/docs/44/en/html/runtime-config-failover.html for more details.

- MASTER and other watchdog nodes are taken into account in failover.
  See
  https://www.pgpool.net/docs/44/en/html/runtime-watchdog-config.html#CONFIG-WATCHDOG-FAILOVER-BEHAVIOR
  for more details.

- Load balance node is not taken into account in streaming replication mode.
  See note #2 in https://www.pgpool.net/docs/44/en/html/runtime-config-failover.html for more details.

> 
> *Thanks*
> *Gopi*
> 
> 
> On Tue, May 16, 2023 at 10:56 AM Gopikrishnan <nksgopikrishnan at gmail.com>
> wrote:
> 
>> Thank you for your response!
>>
>> Can you also kindly clarify this:
>>  During failover, Is the '*primary*' or '*MASTER*' or '
>> *load_balance_node* failure taken into account?
>>
>> *Thanks*
>> *Gopi*
>>
>>
>> On Tue, May 16, 2023 at 10:26 AM Bo Peng <pengbo at sraoss.co.jp> wrote:
>>
>>> Hi,
>>>
>>> > Regd. which I couldn't gather much from the docs at:
>>> > https://www.pgpool.net/docs/pgpool-II-4.0.4/
>>> >
>>> >
>>> >    1. primary
>>> >    2. MASTER
>>> >    3. load_balance_node
>>>
>>> 1. primary:
>>>    It is the role of *PostgreSQL* nodes.
>>>    The role of PostgreSQL nodes can be primary or standby
>>>    in streaming replication mode.
>>>
>>> 2. MASTER:
>>>    I think you are asking the status name shown
>>>    in the result of pcp_watchdog_info command.
>>>    The status name MASTER/STANDBY is the role of *Pgpool-II* node.
>>>    If you have multiple Pgpool-II nodes using watchdog feature,
>>>    one of them will be elected as MASTER and others will be STANDBY.
>>>
>>> 3. load_balance_node
>>>    Pgpool-II supports for load balancing of SELECT queries.
>>>    When a session starts, pgpool will randomly elect one PostgreSQL node
>>>    as the load balance node (load_balance_node) and routes SELECTs to
>>>    the load_balance_node.
>>>
>>> On Mon, 15 May 2023 15:54:53 +0530
>>> Gopikrishnan <nksgopikrishnan at gmail.com> wrote:
>>>
>>> > Hi folks,
>>> >
>>> > Need your help in clearing a confusion regd. basic terminologies:
>>> >
>>> > Regd. which I couldn't gather much from the docs at:
>>> > https://www.pgpool.net/docs/pgpool-II-4.0.4/
>>> >
>>> >
>>> >    1. primary
>>> >    2. MASTER
>>> >    3. load_balance_node
>>> >
>>> > I have a pgpool with 3 nodes. I have a confusion between the pgpool
>>> primary and
>>> > the watchdog *MASTER* and the `*load_balance_node*`
>>> >
>>> > In my setup, the pgpool primary node (according to show pool_nodes
>>> output)
>>> > is : *10.108.104.31, *and also the node designated *load_balance_node*.
>>> >
>>> > While the MASTER node, according to pgpool watchdog, is:
>>> *10.108.104.32* And
>>> > the *cluster virtual IP* is attached to this node.
>>> >
>>> > I have the following questions in mind:
>>> >
>>> >
>>> >    1. What do the terms '*primary*' and '*MASTER*' and
>>> *load_balance_node*
>>> >    mean?
>>> >    2. During failover, Is the '*primary*' or '*MASTER*' or '
>>> >    *load_balance_node* failure taken into account?
>>> >    3. What is the difference (or any relation) between the '*primary*'
>>> and '
>>> >    *MASTER*' and *load_balance_node* nodes?
>>> >    4. If the cluster virtual IP is on the non '*primary*' node,
>>> wouldn't it
>>> >    affect the performance of the HA setup?
>>> >
>>> > Below are the outputs for *show pool_nodes *and* watchdog *info:
>>> >
>>> > psql -h delegateIP -p 9999 -U pgpool postgres -c "show pool_nodes"
>>> > Password for user pgpool:
>>> >  node_id |   hostname    | port | status | lb_weight |  role   |
>>> > select_cnt | load_balance_node | replication_delay |
>>> >
>>> last_status_change---------+---------------+------+--------+-----------+---------+------------+-------------------+-------------------+---------------------
>>> >  0       | 10.108.104.31 | 5432 | up     | 0.333333  | primary | 0
>>> >      | true              | 0                 | 2023-05-12 12:07:13
>>> >  1       | 10.108.104.32 | 5432 | up     | 0.333333  | standby | 0
>>> >      | false             | 0                 | 2023-05-12 12:07:13
>>> >  2       | 10.108.104.33 | 5432 | down   | 0.333333  | standby | 0
>>> >      | false             | 0                 | 2023-05-12 12:07:13
>>> > (3 rows)
>>> >
>>> > And the Watchdog info:
>>> >
>>> > pcp_watchdog_info -h localhost -vw -U pgpool
>>> > Watchdog Cluster Information
>>> > Total Nodes          : 3
>>> > Remote Nodes         : 2
>>> > Quorum state         : QUORUM EXIST
>>> > Alive Remote Nodes   : 2
>>> > VIP up on local node : NO
>>> > Master Node Name     : vm-104-32.eng.dev.com:9999 Linux
>>> vm-104-32.eng.dev.com
>>> > Master Host Name     : 10.108.104.32
>>> >
>>> > Watchdog Node Information
>>> > Node Name      : vm-104-31.eng.dev.com:9999 Linux vm-104-31.eng.dev.com
>>> > Host Name      : vm-104-31.eng.dev.com
>>> > Delegate IP    : 10.108.104.34
>>> > Pgpool port    : 9999
>>> > Watchdog port  : 9000
>>> > Node priority  : 1
>>> > Status         : 7
>>> > Status Name    : STANDBY
>>> >
>>> > Node Name      : vm-104-32.eng.dev.com:9999 Linux vm-104-32.eng.dev.com
>>> > Host Name      : 10.108.104.32
>>> > Delegate IP    : 10.108.104.34
>>> > Pgpool port    : 9999
>>> > Watchdog port  : 9000
>>> > Node priority  : 1
>>> > Status         : 4
>>> > Status Name    : MASTER
>>> >
>>> > Node Name      : vm-104-33.eng.dev.com:9999 Linux vm-104-33.eng.dev.com
>>> > Host Name      : 10.108.104.33
>>> > Delegate IP    : 10.108.104.34
>>> > Pgpool port    : 9999
>>> > Watchdog port  : 9000
>>> > Node priority  : 1
>>> > Status         : 7
>>> > Status Name    : STANDBY
>>> >
>>> >
>>> >
>>> > *Thanks & Regards*
>>> > *Gopi*
>>>
>>>
>>> --
>>> Bo Peng <pengbo at sraoss.co.jp>
>>> SRA OSS LLC
>>> https://www.sraoss.co.jp/
>>>
>>


More information about the pgpool-general mailing list