Bug #15200: A realm pull fails if radosgw process on the master zone is not restarted. - rgw - Ceph

Actions

Copy link

Bug #15200

closed

A realm pull fails if radosgw process on the master zone is not restarted.

Added by Shilpa MJ about 8 years ago. Updated about 7 years ago.

Status:

Resolved

Priority:

Normal

Assignee:

Casey Bodley

Target version:

% Done:

Source:

other

Tags:

Backport:

Regression:

Severity:

3 - minor

Reviewed:

Affected Versions:

Ceph - v10.0.4

ceph-qa-suite:

rgw

Pull request ID:

Crash signature (v1):

Crash signature (v2):

Description

With the new rgw multisite configuration, while doing a realm pull from a master zone the following error appears:

$ radosgw-admin realm pull --url=http://rgw1:7480 --access-key=$access-key --secret=$secret
7f0d3bc9e780 20 sending request to http://rgw1:7480/admin/realm
7f0d3bc9e780 10 receive_http_header
7f0d3bc9e780 10 received header:HTTP/1.1 403 Forbidden
7f0d3bc9e780 10 receive_http_header
7f0d3bc9e780 10 received header:x-amz-request-id: tx000000000000000000047-0056ebee33-100a-default
7f0d3bc9e780 10 receive_http_header
7f0d3bc9e780 10 received header:Content-Length: 123
request failed: (13) Permission denied

If the radosgw process is restarted on the master zone rgw1, the realm pull is successful. The expectation is that the radosgw need not be restarted.

Actions

Copy link

Updated by Abhishek Lekshmanan about 8 years ago

I've also run into this, if I understand correctly,this is something like, the radosgw instance still runs in the default/prev. zone and tries to authenticate the user by checking for the user in the already existing default users.uid pool instead of the newly created zone.

Actions

Copy link