Project

General

Profile

Bug #44180

cephadm: missing describe_service call crashes the MGR when accessing Dashboard

Added by Kiefer Chang about 4 years ago. Updated about 4 years ago.

Status:
Resolved
Priority:
Normal
Assignee:
-
Category:
cephadm
Target version:
-
% Done:

0%

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

Description

How to reproduce:

1. Enable cephadm
bin/ceph mgr module enable cephadm
bin/ceph orch set backend cephadm

2. Access Dashboard and log in. mgr daemon crashes on the landing page (segfault).

After some diggings, turn out that's because on the landing page we make requests to `/api/health/minimal`, and it calls describe_service in orchestrator backend for iSCSI services.

Call path:
https://github.com/ceph/ceph/blob/ae5d2c6ab23bc84fd50b3289bfd52941822f0d1b/src/pybind/mgr/dashboard/controllers/health.py#L126
https://github.com/ceph/ceph/blob/ae5d2c6ab23bc84fd50b3289bfd52941822f0d1b/src/pybind/mgr/dashboard/services/iscsi_config.py#L125
https://github.com/ceph/ceph/blob/ae5d2c6ab23bc84fd50b3289bfd52941822f0d1b/src/pybind/mgr/dashboard/services/iscsi_config.py#L83

The gdb says the mgr process is terminated becasue segfault, not sure why an unimplemented method causes this.

Will the call `describe_service` be added back in the future?

History

#1 Updated by Sebastian Wagner about 4 years ago

+1 for adding mypy to mgr/dashboard.

#2 Updated by Kiefer Chang about 4 years ago

  • Status changed from New to Resolved

#3 Updated by Kiefer Chang about 4 years ago

  • Pull request ID set to 33359

Also available in: Atom PDF