Project

General

Profile

Actions

Bug #65721

open

[MON] Connection Scores: peers become dead after ~5mins, However quorum seems fine

Added by Kamoltat (Junior) Sirivadhna 21 days ago. Updated 21 days ago.

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

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Component(RADOS):
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

All ranks reports that everyone is alive and well

{
    "rank": 0,
    "epoch": 17,
    "version": 0,
    "half_life": 43200,
    "persist_interval": 10,
    "reports": [
        {
            "rank": 0,
            "epoch": 17,
            "version": 0,
            "peer_scores": [
                {
                    "peer_rank": 1,
                    "peer_score": 0.96207665492399386,
                    "peer_alive": true
                },
                {
                    "peer_rank": 2,
                    "peer_score": 0.95951592541247566,
                    "peer_alive": true
                }
            ]
        },
        {
            "rank": 1,
            "epoch": 15,
            "version": 12,
            "peer_scores": [
                {
                    "peer_rank": 0,
                    "peer_score": 0.95957892105899378,
                    "peer_alive": true
                },
                {
                    "peer_rank": 2,
                    "peer_score": 0.95973276624026949,
                    "peer_alive": true
                }
            ]
        },
        {
            "rank": 2,
            "epoch": 15,
            "version": 12,
            "peer_scores": [
                {
                    "peer_rank": 0,
                    "peer_score": 0.95960875128134437,
                    "peer_alive": true
                },
                {
                    "peer_rank": 1,
                    "peer_score": 0.95980374677352776,
                    "peer_alive": true
                }
            ]
        }
    ]
}

After ~5mins rank 0 starts dead pinging ranks 1 & 2

Rank 1 seems to be getting the pings from ranks 0 & 2.

Rank 2 starts dead-pinging rank 0.


{
    "rank": 0,
    "epoch": 22,
    "version": 35175,
    "half_life": 43200,
    "persist_interval": 10,
    "reports": [
        {
            "rank": 0,
            "epoch": 22,
            "version": 35175,
            "peer_scores": [
                {
                    "peer_rank": 1,
                    "peer_score": 0.17510481213537046,
                    "peer_alive": false
                },
                {
                    "peer_rank": 2,
                    "peer_score": 0.14051827066514111,
                    "peer_alive": false
                }
            ]
        },
        {
            "rank": 1,
            "epoch": 22,
            "version": 3532,
            "peer_scores": [
                {
                    "peer_rank": 0,
                    "peer_score": 0.76423188717530466,
                    "peer_alive": true
                },
                {
                    "peer_rank": 2,
                    "peer_score": 0.76466662693832665,
                    "peer_alive": true
                }
            ]
        },
        {
            "rank": 2,
            "epoch": 22,
            "version": 2565,
            "peer_scores": [
                {
                    "peer_rank": 0,
                    "peer_score": 0.71389248169284003,
                    "peer_alive": false
                },
                {
                    "peer_rank": 1,
                    "peer_score": 0.76385711696841496,
                    "peer_alive": true
                }
            ]
        }
    ]
Actions #1

Updated by Kamoltat (Junior) Sirivadhna 21 days ago

  • Subject changed from [MON] Connection Scores: peers become dead after ~5mins to [MON] Connection Scores: peers become dead after ~5mins, However quorum seems fine
Actions

Also available in: Atom PDF