Bug #51538
closedmultisite sync not syncing any data via HTTPS endpoints
0%
Description
I setup two clusters with RADOSGW and setup a multi-site sync, just as described in https://docs.ceph.com/en/octopus/radosgw/multisite/.
Things went well and meta-data was in sync just seconds after startup of the gateways.
But there was no sync of data shards / bucket data happening which lead me to post to ML (https://lists.ceph.io/hyperkitty/list/ceph-users@ceph.io/thread/VP7BOHRGIEIQAMHJIIZWSR73PLLQJCYO) about possible clues on the error:
2021-06-25T16:20:29.167+0000 7f87c2aec080 0 RGW-SYNC:data:init_data_sync_status: ERROR: failed to read remote data log shards
I then found this other post reporting a similar problem solved by switching to HTTP on the endpoints: https://lists.ceph.io/hyperkitty/list/ceph-users@ceph.io/thread/637W3R5M4PKGHSDR2AFGKG6HEEUL6C5L/
I then simply added a seconds port to the civetweb
rgw frontends = "civetweb port=7481+7480s ssl ...."
and reconfigured the endpoints on my zones. Data sync then stated immediately and things are just working.
I suspect the rgw data sync client has some sort of bug like mishandling of the endpoint URL to cause this?
Updated by Christian Rohmann over 2 years ago
I narrowed this down to some config issue on the reverse proxy / loadbalancer and this issue can be closed.