multi-zone replication doesn't work
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Ceph RADOS Gateway Charm |
Fix Released
|
High
|
James Page | ||
ceph (Ubuntu) |
Invalid
|
Undecided
|
Unassigned |
Bug Description
I deploy the following bundle
https:/
(tried both bionic-ussuri and focal-distro origins).
There are two radosgw instances, each is backed by its own Ceph cluster.
Each RGW pull info correctly, however the sync status on master:
root@juju-
realm 2ce9452b-
zonegroup f68b626c-
zone c2d6f2e0-
metadata sync no sync (zone is master)
2021-03-
data sync source: a2c19e5e-
On slave:
root@juju-
realm 2ce9452b-
zonegroup f68b626c-
zone a2c19e5e-
metadata sync preparing for full sync
data sync source: c2d6f2e0-
I see the system use on the master but don't see on the slave.
The zones are configured correctly on both, and the slave can pull the realm successfully.
One suspicious thing, on the slave there is a complaint in the log:
2021-03-
This is the access key of the system user used for replication, and it is correct (that's the one created on the master.
Changed in charm-ceph-radosgw: | |
assignee: | nobody → James Page (james-page) |
status: | Triaged → In Progress |
Changed in charm-ceph-radosgw: | |
milestone: | none → 21.10 |
Changed in charm-ceph-radosgw: | |
status: | Fix Committed → Fix Released |
I can confirm I see the same behaviour using a slightly adapted version of the bundle.