PMM3.7 - Getting error "refused to connect" on all panels after upgrade to 3.7.0

Percona Monitoring and Management (PMM) PMM 3.x
1

Hello dear Percona team,

after upgrading PMM from 3.5.0 to 3.7.0, we started to observed an error on some pages. Not the entire page, actually, but on frames within the page. The console loads without any issues, and we can see the side menu (and interact with it), but internal frames are returning the error “ refused to connect“.

The container logs seems to not show any issues:

Running as UID 1000
Checking /usr/share/pmm-server directory structure...
Generating self-signed certificates for nginx...
Checking nginx configuration...
2026/04/15 09:22:21 [warn] 17#17: "ssl_stapling" ignored, issuer certificate not found for certificate "/srv/nginx/wildcard_2026.pem"
nginx: [warn] "ssl_stapling" ignored, issuer certificate not found for certificate "/srv/nginx/wildcard_2026.pem"
nginx: the configuration file /etc/nginx/nginx.conf syntax is ok
nginx: configuration file /etc/nginx/nginx.conf test is successful
time="2026-04-15T09:22:21.735+00:00" level=info msg="pmm.ini configuration has been updated."
2026-04-15 09:22:21,938 INFO Included extra file "/etc/supervisord.d/grafana.ini" during parsing
2026-04-15 09:22:21,938 INFO Included extra file "/etc/supervisord.d/pmm.ini" during parsing
2026-04-15 09:22:21,938 INFO Included extra file "/etc/supervisord.d/qan-api2.ini" during parsing
2026-04-15 09:22:21,938 INFO Included extra file "/etc/supervisord.d/supervisord.ini" during parsing
2026-04-15 09:22:21,938 INFO Included extra file "/etc/supervisord.d/victoriametrics.ini" during parsing
2026-04-15 09:22:21,938 INFO Included extra file "/etc/supervisord.d/vmalert.ini" during parsing
2026-04-15 09:22:21,938 INFO Included extra file "/etc/supervisord.d/vmproxy.ini" during parsing
2026-04-15 09:22:21,941 INFO RPC interface 'supervisor' initialized
2026-04-15 09:22:21,942 INFO supervisord started with pid 1
2026-04-15 09:22:22,945 INFO spawned: 'pmm-init' with pid 23
2026-04-15 09:22:22,947 INFO spawned: 'postgresql' with pid 24
2026-04-15 09:22:22,949 INFO spawned: 'clickhouse' with pid 25
2026-04-15 09:22:22,951 INFO spawned: 'grafana' with pid 26
2026-04-15 09:22:22,953 INFO spawned: 'nginx' with pid 27
2026-04-15 09:22:22,955 INFO spawned: 'victoriametrics' with pid 30
2026-04-15 09:22:22,957 INFO spawned: 'vmalert' with pid 31
2026-04-15 09:22:22,959 INFO spawned: 'vmproxy' with pid 32
2026-04-15 09:22:22,961 INFO spawned: 'qan-api2' with pid 34
2026-04-15 09:22:22,963 INFO spawned: 'pmm-managed' with pid 42
2026-04-15 09:22:22,974 INFO exited: qan-api2 (exit status 1; not expected)
2026-04-15 09:22:23,943 INFO success: pmm-init entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2026-04-15 09:22:23,946 INFO success: postgresql entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2026-04-15 09:22:23,948 INFO success: clickhouse entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2026-04-15 09:22:23,950 INFO success: grafana entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2026-04-15 09:22:23,951 INFO success: nginx entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2026-04-15 09:22:23,954 INFO success: victoriametrics entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2026-04-15 09:22:23,956 INFO success: vmalert entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2026-04-15 09:22:23,958 INFO success: vmproxy entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2026-04-15 09:22:23,964 INFO success: pmm-managed entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2026-04-15 09:22:23,981 INFO spawned: 'qan-api2' with pid 818
2026-04-15 09:22:24,024 INFO exited: qan-api2 (exit status 1; not expected)
2026-04-15 09:22:24,448 INFO spawned: 'pmm-agent' with pid 881
2026-04-15 09:22:25,447 INFO success: pmm-agent entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2026-04-15 09:22:26,037 INFO spawned: 'qan-api2' with pid 1048
2026-04-15 09:22:26,786 INFO exited: pmm-init (exit status 0; expected)
2026-04-15 09:22:27,037 INFO success: qan-api2 entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)

Within the container, checking the “grafana.log“ I can see a lot of these messages (not sure if it’s related):

logger=context userId=0 orgId=0 uname= t=2026-04-15T09:55:37.842373884Z level=info msg="Request Completed" method=GET path=/api/live/ws status=401 remote_addr=127.0.0.1 time_ms=0 duration=662.336µs size=105 referer= handler=/api/live/ws status_source=server errorReason=Unauthorized errorMessageID=session.token.rotate error="token needs to be rotated"

Container is showing as healthy and I can access it normally, so not sure what changed from PMM3.5 to PMM3.7 to cause this.

Important to mention that last year we upgraded PMM from 3.3.1 to 3.5 in the same way without any issues.

Do you have any idea why we are facing this issue?

Thanks for your assistance.

2

HI @FabricioPedrosoJorge, we had a report of the same Grafana log in PMM 3.2.0 - PMM-14121. Unfortunately, it was not reproducible by us.

I also just tried upgrading from 3.5.0 to 3.7.0, but didn’t encounter this issue.

Do you use any external authentication with Grafana?

Also, as a quick check, could you clear the browser’s cookies (or use a private window) and see if it fixes the issue?

3

@FabricioPedrosoJorge

Also, in the latest PMM 3.7, the Percona Platform feature is removed in case you using or have some dependency earlier.

> Percona Platform connectivity removed

As announced in PMM 3.5.0, Percona Platform services are now discontinued.

With PMM 3.7.0, there’s no longer any connection to Percona Platform in the UI or API. Since your advisors and alert templates now ship directly with PMM, everything will continue working exactly as before, you just won’t see any Platform-related options anymore.

This also means you can no longer sign into PMM using your Percona Account. If you used this login method, switch to a different authentication method such as basic auth, LDAP, OAuth, or SAML.

There are some version-related and other changes added, which you can read in more detail under the release note - PMM 3.7.0 (2026-04-01) - Percona Monitoring and Management in case you are using any specific external changes.

Are those some customized dashbaord where you seeing the issues ? Can you share the exact screenshot of the dashiboard or panel where you getting the errors for a better clarity ?

Well, if you can share the PMM Server upgrade and other component logs [/srv/log/*.log
] from the container, we may be able to check this in detail.