Ceph

12/22/2021

12:41 PM rgw Bug #53698 (Won't Fix - EOL): a slow reader of a large object receives corrupt object contents from rgw with civetweb frontend

first, the real-life scenario:
we have been copying a number of large objects from swift interface of rgw of a mim... Jaka Močnik

11/26/2021

09:15 AM rgw Bug #53384: tail objects that have already been garbage collected remain in the gc queue forever

fwiw, 12 hours after setting max_time and period on all rgw instances to 1hr, the gc queue is empty. thanks again. Jaka Močnik

11/25/2021

12:28 PM rgw Bug #53384: tail objects that have already been garbage collected remain in the gc queue forever

I understand. however, as stated, the current behaviour is a bug as it will result in the gc queue never draining if ... Jaka Močnik

09:59 AM rgw Bug #53384: tail objects that have already been garbage collected remain in the gc queue forever

thanks for the insight, pritha.
I was suspecting the fact that large deletes take over 5mins time could be related... Jaka Močnik

11/24/2021

04:03 PM rgw Bug #53384: tail objects that have already been garbage collected remain in the gc queue forever

attaching complete log files - save the ones that containt HTTP_X_AUTH_TOKEN - of said rgw instance for some 40 minut... Jaka Močnik

02:21 PM rgw Bug #53384: tail objects that have already been garbage collected remain in the gc queue forever

explicitly set rgw-related config options:
rgw_override_bucket_index_max_shards = 16
rgw_max_put_size = 109951162... Jaka Močnik

02:13 PM rgw Bug #53384 (Triaged): tail objects that have already been garbage collected remain in the gc queue forever

running an octopus cluster (upgraded from nautilus a few months ago) of some 0.5PB capacity. it is used exclusively a... Jaka Močnik