Project

General

Profile

Bug #24389

ceph dashboard set-login-credentials results in an error

Added by Michael Sudnick almost 6 years ago. Updated almost 3 years ago.

Status:
Resolved
Priority:
Normal
Assignee:
Category:
General
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

I get
Cluster fully updated to Mimic 13.2.0 from the more recent luminous release by following the release notes upgrade procedure. All nodes are running CentOS 7.5 with OSDs, MONs, MGRs, and MDSs all co-located.

Error EIO: Module 'dashboard' has experienced an error and cannot handle commands: cannot import name DependencyWarning

When I try to set a username and password.

My mgr daemons all have the following caps: mds "allow *" mon "allow profile mgr" osd "allow *"

Creating a self-signed certificate worked, however trying to access the webpage results in firefox being unable to establish a connection. I have tried disabling and re-enabling the dashboard module to no avail.

Perhaps a dependency didn't get pulled in by yum?

mon-log.tar.bz2 - Log from monitor (259 KB) Michael Sudnick, 06/02/2018 08:19 PM

ceph logs.tar.bz2 (95.7 KB) Michael Sudnick, 06/03/2018 01:36 PM

History

#1 Updated by Lenz Grimmer almost 6 years ago

  • Status changed from New to Need More Info

Michael Sudnick wrote:

Error EIO: Module 'dashboard' has experienced an error and cannot handle commands: cannot import name DependencyWarning

When I try to set a username and password.

Could you please take a look at the MON's log files and check if you can find a more detailed Python stack trace? For some reason, the Dashboard does not even start up - the log should provide more insight into what's going on.

#2 Updated by Michael Sudnick almost 6 years ago

Added a log file, set mon debug to 20/20. Doesn't seem to give anything illuminating, at least to me. Any suggestion on how to find a more detailed stack trace?

#3 Updated by Lenz Grimmer almost 6 years ago

Michael Sudnick wrote:

Added a log file, set mon debug to 20/20. Doesn't seem to give anything illuminating, at least to me. Any suggestion on how to find a more detailed stack trace?

This seems to be the log file of a CephFS Metadata Server (MDS), not from the Ceph Manager process. Please double check that you check the appropriate log file, which usually has the string "mgr" in its file name. It's the Ceph Manager manages the dashboard plugin, so this is where an error should be logged.

Please see http://docs.ceph.com/docs/mimic/rados/troubleshooting/log-and-debug/ for more details on how to enable logging and debugging.

#4 Updated by Michael Sudnick almost 6 years ago

Sorry about that, I've attached a tar of the active mds log, and the active mon log. I set logging to both to 20/20, but maybe I did that wrong, I'm still not seeing anything informative.

#5 Updated by Ricardo Dias almost 6 years ago

Can you check if you have the "python-requests" package installed in your system?

#6 Updated by Michael Sudnick almost 6 years ago

python-requests-2.6.0-1.el7_1 is installed.

#7 Updated by Ricardo Dias almost 6 years ago

I was able to reproduce the issue by managing to install a newer version of the "requests" python package using "pip" with an older version of "urllib3" python package using "yum".

The solution should be to guarantee that the urllib3 version matches what's required by the requests package.

Here's a sequence of commands that might solve your issue:

$ pip uninstall urllib3
$ pip install urllib3 --upgrade

#8 Updated by Michael Sudnick almost 6 years ago

That fixed the issue, thank you!

#9 Updated by Lenz Grimmer almost 6 years ago

  • Status changed from Need More Info to Resolved

#10 Updated by Ernesto Puerta almost 3 years ago

  • Project changed from mgr to Dashboard
  • Category changed from 132 to General

Also available in: Atom PDF