digitalocean / ceph_exporter

Prometheus exporter that scrapes meta information about a ceph cluster.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

jewel not collect monitor data

proffust opened this issue · comments

Hello!
root@ceph-osd-mon02:~# ceph --version
ceph version 10.2.10 (5dc1e4c05cb68dbf62ae6fce3f0700e4654fdbbe)
Using exporter in docker image
docker run -v /etc/ceph:/etc/ceph -d --net=host digitalocean/ceph_exporter
OSD metrics collect normal, but no monitor metrics. In logs :
root@ceph-osd-mon02:~# docker logs -f a70ac49a79d3
2018/07/19 16:29:19 Starting ceph exporter on ":9128"
2018/07/19 16:29:29 failed collecting monitor metrics: rados: Invalid argument
2018/07/19 16:41:56 failed collecting monitor metrics: rados: Invalid argument

ceph.conf - https://pastebin.com/sYPgqPSR

metoo

I solved this problem using correct software version, maybe it's same problem

https://github.com/digitalocean/ceph_exporter/releases/tag/2.0.0-jewel

me too

commit 6f29075
Author: Vaibhav Bhembre vaibhav@digitalocean.com
Date: Sat Jun 23 18:51:06 2018 -0400

0.Remove commit from monitor.go If u use ceph version 0.94 or 10.2
1.This is commit only support version of L +

[xxoo@xxoo ~]# ceph time-sync-status  --format=json |  python -m json.tool
{
    "time_skew_status": {
        "rg1-ceph01": {
            "health": "HEALTH_OK",
            "latency": 0.0,
            "skew": 0.0
        },
        "rg1-ceph02": {
            "health": "HEALTH_OK",
            "latency": 0.00052,
            "skew": 0.0
        },
        "rg1-ceph03": {
            "health": "HEALTH_OK",
            "latency": 0.000695,
            "skew": -0.000163
        }
    },
    "timechecks": {
        "epoch": 42,
        "round": 80862,
        "round_status": "finished"
    }
}

3.go build again
4.back to my vim editor and code...... haha666

I have the same problem with ceph version 0.94.9

Same problem with 10.2.11 cluster.

commented

As mentioned above this seems to have been caused by using the incorrect version of ceph exporter for the cluster. #96 (comment)

We'll soon have support for Nautilus, Octopus, and Pacific clusters with a single version of ceph exporter.