The restriction seems to be due to truncate being used on capped collections.
Dan Gottlieb clarified in the subsequent comment how capped collections do not use WT's truncate to do the deletion. The issue for snapshot reads is that deletes weren't happening at a point-in-time due to not being replicated. SERVER-16049 changed the behavior where capped deletions are now replicated. Furthermore, SERVER-56230 was meant to remove the restriction and thus enable running the dbHash command at a point-in-time in a snapshot read. CC gregory.wlodarek@mongodb.com
pierlauro.sciarelli@mongodb.com, can you clarify the ask for Cluster Scalability? As far as I understand this limitation for capped collections and snapshot reads is artificial and either Catalog and Routing or Storage Execution would be responsible for removing it.
|