lakeFS is an open-source data version control that transforms your object storage to Git-like repositories..

lakeFS

Hi, I have a question about GC: If I only call `getPhysicalAddress` and am writing a file through the S3 interface, and GC is triggered before `linkPhysicalAddress` is called, the S3 object will be collected but not marked as active. Will this cause a false GC?

Hi Yuhao,
Physical addresses are valid for many minutes, but not for days.  You should not keep a physical address for as long as the GC period.
If you have an explicit need for such a long period of object creation - let's discuss!

To be clear: we _do_ verify the link is still valid (see the aptly-named <https://github.com/treeverse/lakeFS/blob/a8cd3b4f33ec4599834a6aecd3704db5f71a32a5/pkg/catalog/catalog.go#L2920-L2922|VerifyLinkAddress>), so you will receive an error message when you try to do this.

<@U01867G8AA0> Oh, that makes sense. So within six hours after `getPhysicalAddress` is called, the PhysicalAddress will be marked as active, so it will not be cleaned up by GC.

To clarify on <@U01867G8AA0> point. We have a 6 hour grace in GC which correlates to the getPhysicalAddresses expiry. Which means GC will not delete objects that were uploaded less than 6 hours ago