Nos stats problem (but not the know one I guess)

cgasser

I have upgraded from 8.2.1 to 8.3. No I cant see any stats anymore. But I get a lot of these messages:

{
  "id": "0mfjnqn6i",
  "properties": {
    "method": "vm.stats",
    "params": {
      "id": "2dd9279e-1f06-0151-6b9a-1ebf29d48edc"
    },
    "name": "API call: vm.stats",
    "userId": "7a3187b0-312e-43f9-9870-cfd591078f02",
    "type": "api.call"
  },
  "start": 1757852076761,
  "status": "failure",
  "updatedAt": 1757852076845,
  "end": 1757852076845,
  "result": {
    "errno": -111,
    "code": "ECONNREFUSED",
    "syscall": "connect",
    "address": "127.0.0.1",
    "port": 443,
    "message": "connect ECONNREFUSED 127.0.0.1:443",
    "name": "Error",
    "stack": "Error: connect ECONNREFUSED 127.0.0.1:443\n    at TCPConnectWrap.afterConnect [as oncomplete] (node:net:1555:16)\n    at TCPConnectWrap.callbackTrampoline (node:internal/async_hooks:128:17)"
  }
}

To me it looks like XOA tries to load the data from 127.0.0.1.
I have another pool connected to that XOA. Also 8.3. No issues. Therefore I assume it is not my XOA that causes that problem. I also update to the latest version via git pull. No change.

Anyone and idea why this is?

BR
C

cgasser

Update: I have 4 hosts in that pool. For one host, I can see stats. But just for that one host only. Not sure if that helps to point to the right directrion.

BR
C

olivierlambert

@cgasser said in Nos stats problem (but not the know one I guess):

connect ECONNREFUSED 127.0.0.1:443

That's the issue. XO needs to have access to all hosts management IPs, not just the master. I assume you made a tunnel or something specific to make XO connecting to the master. It's not related to 8.2 or 8.3, it's an architecture thing on pools since XCP exists

cgasser

@olivierlambert Hi! No, that is not the case. All servers are in the same rack. Connected to the same switch. Same network (VLAN, IP Range).

XOA runs on the master server. No stats for that server. But stats for another server are ok (just one), all the others are not working.

C

olivierlambert

Then, there is one machine thinking its management IP is 127.0.0.1

cgasser

@olivierlambert Right. Any idea on how to check and change that? Management interfaces seems to be OK on the hosts. I can access and manage all servers. It seems it is just the stats that are not working. C

olivierlambert

I would print the result of the command xe host-param-list uuid=<problematic host UUID> to see where this localhost IP is set somlewhere.

cgasser

@olivierlambert Nothing with 127.0.0.1. Also hostname ( RO): is set correct. I can find some localhost entries. But for license server, etc. C

olivierlambert

Let's try to grep 127.0.0.1 in the XAPI DB file then

cgasser

@olivierlambert Would you mind giving me a hint on where to find that file? C

cgasser

Folder /var/lib/xcp/ file xapi? C

cgasser

If it is the state.db file, then there is no mention of 127.0.0.1.

grep "127.0.0.1" state.db brings nothing no result. C