You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
(was asked to note this as part of a phone conversation later today; this feature request is just to document a concern/usecase)
This feature request is to make it easier to individually backup and restore individual repos within a multi-repo DVID system.
Right now, our tracing DVID has many segmentations in it, and probably always will unless we change policy.
I take nightly backups of our tracing DVID (by stopping the service, and tarring up the directory our DVID service). Actually doing restores of this would be very painful (because the repos and the instances inside them are all stored in the same backend store, and because there are not easy ways to pull instances out of the backup and push them back into the server as replacements, even if I start up a second DVID server based on the backup). I likely would need to restore the entire server state rather than just the repo that I want, meaning deciding to backup and delete a seldom-used repo becomes a high-stakes decision.
One possible solution would be for each repo to naturally land in a separate backend storage by default (we programmatically make new repos sometimes and shouldn't need to create new config entries for this), and for them to be file based; it would work just as well for this to have separate dump and restore commands on a per-repo basis (in the way that one can dump and restore schemas in Postgres).
The text was updated successfully, but these errors were encountered:
(was asked to note this as part of a phone conversation later today; this feature request is just to document a concern/usecase)
This feature request is to make it easier to individually backup and restore individual repos within a multi-repo DVID system.
Right now, our tracing DVID has many segmentations in it, and probably always will unless we change policy.
I take nightly backups of our tracing DVID (by stopping the service, and tarring up the directory our DVID service). Actually doing restores of this would be very painful (because the repos and the instances inside them are all stored in the same backend store, and because there are not easy ways to pull instances out of the backup and push them back into the server as replacements, even if I start up a second DVID server based on the backup). I likely would need to restore the entire server state rather than just the repo that I want, meaning deciding to backup and delete a seldom-used repo becomes a high-stakes decision.
One possible solution would be for each repo to naturally land in a separate backend storage by default (we programmatically make new repos sometimes and shouldn't need to create new config entries for this), and for them to be file based; it would work just as well for this to have separate dump and restore commands on a per-repo basis (in the way that one can dump and restore schemas in Postgres).
The text was updated successfully, but these errors were encountered: