Project

General

Profile

Actions

Support #52881

open

Filtered out host node3.foo.com: does not belong to mon public_network ()

Added by Ralph Soika over 2 years ago. Updated over 2 years ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
-
Target version:
-
% Done:

0%

Tags:
logging
Reviewed:
Affected Versions:
Component(RADOS):
OSD
Pull request ID:

Description

I am running a Ceph Pacific cluster ( version 16.2.6) consisting of 3 nodes with public Internet Addresses. I also have a private network for OSD replication.

The node IPs looks something like this:

127.0.1.1 node3
10.0.0.4 node3
23.88.xxx.yyy node3.foo.com

The cluster log generates each minute an info message like this one:

Filtered out host node3: does not belong to mon public_network ()

My public network is the Internet (0.0.0.0 I guess). And I can't see what this message means to me. Everything works fine in my cluster. It's just the message.

When I look into the OSD Metadata of my hosts, all values seems also to be fine for me:

back_addr [v2:10.0.0.4:6802/1903607507,v1:10.0.0.4:6803/1903607507]
front_addr [v2:23.88.xxx.yyy:6800/1903607507,v1:23.88.xxx.yyy:6801/1903607507]
hb_back_addr[v2:10.0.0.4:6806/1903607507,v1:10.0.0.4:6807/1903607507]
hb_front_addr[v2:23.88.xxx.yyy:6804/1903607507,v1:23.88.xxx.yyy:6805/1903607507]
hostname node3

Can it bee that this message is really just an info? But for what reason?

I found some code snippet (maybe responsible) here:

https://fossies.org/linux/ceph/src/pybind/mgr/cephadm/serve.py

....
671 def matches_network(host):
672 # type: (str) -> bool
673 # make sure we have 1 or more IPs for any of those networks on that
674 # host
675 for network in public_networks:
676 if len(self.mgr.cache.networks[host].get(network, [])) > 0:
677 return True
678 self.log.info(
679 f"Filtered out host {host}: does not belong to mon public_network"
680 f" ({','.join(public_networks)})"
681 )
682 return False
....

But I can't see why the message is generated.

Can I provide more information?

Actions #1

Updated by Ralph Soika over 2 years ago

I can now answer my question by myself. It was a missconfiguration.

After I entered the unmanaged mode with

$ sudo ceph orch apply mon --unmanaged

and added all my monitors, I finally switched back to the managed mode with

$ sudo ceph orch apply mon 3

And this caused the log messages. I guess ceph started to tried now to start monitors but was unable. After switching back to the unmanaged mode all is fine now. So you can close the ticket.
I also posted a tutorial: https://ralph.blog.imixs.com/2021/10/03/ceph-pacific-running-on-debian-11-bullseye/

Actions

Also available in: Atom PDF