On 3/26/22 00:48, Casey Bodley wrote:
multisite test suite:
first, we need a way to issue continuous object writes/deletes
*during* a bucket reshard. python makes this hard
bucket reshard:
* continuous stream of writes/deletes to a bucket in zone A, with one
or more racing 'bucket reshard's on zone A, stop workload and verify
that zone B's checkpoint succeeds
* continuous stream of writes/deletes to a bucket in zone A, with one
or more racing 'bucket reshard's on zone B, stop workload and verify
that zone B's checkpoint succeeds
* as above, but with *failed* bucket reshards using fault injection
similar to https://github.com/ceph/ceph/blob/wip-rgw-multisite-reshard/qa/workunits/rgw/test_rgw_reshard.py#L240
If above can be done, may be we can extend these tests to include
lifecycle as well (as it helped to detect races with dir_suggests). eg.,
(lc_debug_interval & lc_max_workers adjusted)
* add a lc policy to the bucket & then start writes
* set lc policy to delete the objects and in parallel run
'radosgw-admin lc process'
Thanks,
Soumya
_______________________________________________
Dev mailing list -- dev@xxxxxxx
To unsubscribe send an email to dev-leave@xxxxxxx
