I’ve setup PMM on a server, and added a client, which correctly shows up under PMM inventory - but nothing appears in the MYSQL overview, and it doesn’t show on the prometheus scrape list.
Documentation refers to “pmm-admin check-network” but that just errors out with:
wiki mysql # pmm-admin check-network
pmm-admin: error: expected command but got "check-network", try --help
pmm-admin list shows:
wiki mysql # pmm-admin list
Service type Service name Address and port Service ID
MySQL wiki.example.com-mysql 127.0.0.1:3306 /service_id/cbff4c44-2f09-49fb-a0cd-273c4093ee43
Agent type Status Metrics Mode Agent ID Service ID
pmm_agent Connected /agent_id/c49e2a69-33fc-4ddd-9e0f-ec02b7bc5ad8
node_exporter Running push /agent_id/a6ecf473-cf6a-441c-853e-1bc630e96870
mysqld_exporter Running push /agent_id/6b9eb3bd-5989-4dc2-b408-a579e1edb933 /service_id/cbff4c44-2f09-49fb-a0cd-273c4093ee43
mysql_slowlog_agent Running /agent_id/a3706303-1471-4150-8772-1aacbc83ccc3 /service_id/cbff4c44-2f09-49fb-a0cd-273c4093ee43
vmagent Running push /agent_id/8626cb6f-e2d4-4bd1-a1af-43dc6b3f3a72
Running curl from either the server or the client gets a large dump of data:
I’m confused - the documentation is incorrect? You are supposed to add a remove server using that and NOT via pmm-admin on the remote system itself?
To be precise, I have PMM running on a host (pmm.example.com) and I ran the pmm-admin commands on wiki.example.com (another host) - both are VMs running Linux (not a “cloud database” if that matters). That remote server thing seems to be wanting to do remote MySQL commands across the internet? I guess I could do that but it seems … insecure.
It has to be shown on PMM server side according to provided exporter statuses (Running).
Do you see any node metrics on dashboards for host wiki.example.com?
Could you perform pmm-admin summary command and check if any errors are occurred in logs?
It appears the PMM server only sees itself. Attached are the summary ZIPs from both ends here: Box
Looking in the errors indicates it may be an IPv6 issue - the client is trying to communicate with the IPv4 of the server which is a private IP? It should only be using the hostnames.
2021/02/14 13:51:17 [warn] 1139#1139: *10 an upstream response is buffered to a temporary file /var/cache/nginx/proxy_temp/1/00/0000000001 while reading upstream, client: 192.168.88.108, server: _, request: "GET /graph/public/img/login_background_dark.svg HTTP/1.1", upstream: "http://127.0.0.1:3000/public/img/login_background_dark.svg", host: "192.168.88.160", referrer: "http://192.168.88.160/graph/"
We may exclude such issues by specifying IP addresses in a configuration command.
Please notice that after a config command a mysql service has to be added for monitoring again.
e.g. pmm-admin config --server-url="https://admin:K4xu5gDEKQhmsyxW@192.168.88.160" --server-insecure-tls 192.168.88.108 generic myNode