public inbox for [email protected]help / color / mirror / Atom feed
what is causing different pcp_node_info status 4+ messages / 2 participants [nested] [flat]
* what is causing different pcp_node_info status @ 2025-10-17 10:45 Luca Ferrari <[email protected]> 0 siblings, 1 reply; 4+ messages in thread From: Luca Ferrari @ 2025-10-17 10:45 UTC (permalink / raw) To: [email protected] Hi all, I'm using PgPool 4.6.2, PostgreSQL 17.5, three nodes: - pg1 primary - pg2 replica - pg3 replica So far so good, the initial situation of pcp_node_info reported by all the nodes seems consistent: % ssh pg1 'sudo -u postgres pcp_node_info -U pgpool' pg1 5432 2 0.166667 up up primary primary 0 none none 2025-10-16 11:49:45 pg2 5432 2 0.333333 up up standby standby 0 streaming async 2025-10-16 12:02:10 pg3 5432 2 0.500000 up up standby standby 0 streaming async 2025-10-16 11:49:53 % ssh pg2 'sudo -u postgres pcp_node_info -U pgpool' pg1 5432 1 0.166667 waiting up primary primary 0 none none 2025-10-16 11:51:36 pg2 5432 1 0.333333 waiting up standby standby 0 streaming async 2025-10-16 12:02:09 pg3 5432 1 0.500000 waiting up standby standby 0 streaming async 2025-10-16 11:51:36 % ssh pg3 'sudo -u postgres pcp_node_info -U pgpool' pg1 5432 1 0.166667 waiting up primary primary 0 none none 2025-10-16 11:41:22 pg2 5432 1 0.333333 waiting up standby standby 0 streaming async 2025-10-16 12:02:09 pg3 5432 1 0.500000 waiting up standby standby 0 streaming async 2025-10-16 11:49:53 Replicas are streaming, so I reboot pg3 (at 17:30) and the situation appears different: % ssh pg3 'sudo reboot' ^ permalink raw reply [nested|flat] 4+ messages in thread
* Re: what is causing different pcp_node_info status @ 2025-10-23 06:04 Luca Ferrari <[email protected]> parent: Luca Ferrari <[email protected]> 0 siblings, 1 reply; 4+ messages in thread From: Luca Ferrari @ 2025-10-23 06:04 UTC (permalink / raw) To: [email protected] On Fri, Oct 17, 2025 at 12:45 PM Luca Ferrari <[email protected]> wrote: > The problem is that the status of pg3 is not consistent: pg1 reports > as "up up" while other nodes report as "down up". > Please note that the updtae timestamp are in sync, 17:32:45 and hence > the updated seems to have been propagated correctly within pgpool. > Questions are: > - is it safe right now to execute a pcp_attach_node ? > - why is the status not consistent? Anything I should look for within the logs? I am still observing the issue, the best situation I come is all nodes reported as "up up" when queryed on pg1 (primary) and "waiting up" when queried from pg2 and pg3. could it be network latency or a similar issue? Thanks, Luca ^ permalink raw reply [nested|flat] 4+ messages in thread
* Re: what is causing different pcp_node_info status @ 2025-10-29 08:22 Luca Ferrari <[email protected]> parent: Luca Ferrari <[email protected]> 0 siblings, 1 reply; 4+ messages in thread From: Luca Ferrari @ 2025-10-29 08:22 UTC (permalink / raw) To: [email protected] On Thu, Oct 23, 2025 at 8:04 AM Luca Ferrari <[email protected]> wrote: > > I am still observing the issue, the best situation I come is all nodes > reported as "up up" when queryed on pg1 (primary) and "waiting up" > when queried from pg2 and pg3. > could it be network latency or a similar issue? As explained clearly here <https://github.com/pgpool/pgpool2/issues/131#issuecomment-3436197043; the information about the nodes is collected and kept in shared memory. Is there any chance the same process is applied to the pgpool node status? I observed that even a machine reboot does not seem to make pgpool report a status better than "waiting up", evne after minutes, and while replicas are streaming. Restarting then the pgpool service brings the status "up up", this makes me think there is something blocking or delaying pgpool to collect data, and when the service is restarted the node is forced to "dig" the other node status. Something I could investigate on? ^ permalink raw reply [nested|flat] 4+ messages in thread
* Re: what is causing different pcp_node_info status @ 2025-11-04 01:36 Tatsuo Ishii <[email protected]> parent: Luca Ferrari <[email protected]> 0 siblings, 0 replies; 4+ messages in thread From: Tatsuo Ishii @ 2025-11-04 01:36 UTC (permalink / raw) To: [email protected]; +Cc: [email protected] > On Thu, Oct 23, 2025 at 8:04 AM Luca Ferrari <[email protected]> wrote: >> >> I am still observing the issue, the best situation I come is all nodes >> reported as "up up" when queryed on pg1 (primary) and "waiting up" >> when queried from pg2 and pg3. >> could it be network latency or a similar issue? > > As explained clearly here > <https://github.com/pgpool/pgpool2/issues/131#issuecomment-3436197043; > the information about the nodes is collected and kept in shared > memory. > Is there any chance the same process is applied to the pgpool node status? > I observed that even a machine reboot does not seem to make pgpool > report a status better than "waiting up", evne after minutes, and > while replicas are streaming. > Restarting then the pgpool service brings the status "up up", this > makes me think there is something blocking or delaying pgpool to > collect data, and when the service is restarted the node is forced to > "dig" the other node status. > Something I could investigate on? You are misunderstanding what "waiting" means. It means "Node is up. No connections yet." [1] If you are only interested in the node is up or down, you can safely assume that "waiting" == "up". Best regards, [1] https://www.pgpool.net/docs/latest/en/html/pcp-node-info.html -- Tatsuo Ishii SRA OSS K.K. English: http://www.sraoss.co.jp/index_en/ Japanese:http://www.sraoss.co.jp ^ permalink raw reply [nested|flat] 4+ messages in thread
end of thread, other threads:[~2025-11-04 01:36 UTC | newest] Thread overview: 4+ messages (download: mbox.gz follow: Atom feed) -- links below jump to the message on this page -- 2025-10-17 10:45 what is causing different pcp_node_info status Luca Ferrari <[email protected]> 2025-10-23 06:04 ` Luca Ferrari <[email protected]> 2025-10-29 08:22 ` Luca Ferrari <[email protected]> 2025-11-04 01:36 ` Tatsuo Ishii <[email protected]>
This inbox is served by agora; see mirroring instructions for how to clone and mirror all data and code used for this inbox