Hi all,
reading through a CS of an Storage SCP implementation, I came across the following:

"If a duplicate SOP Instance UID is received the system pre-serves the original object but does not report any error."

This made me think about, how most systems actually will handle such a case or some other storage corner cases. Taken for example an image modality that successfully stored and requested storage commitment for an image via C-STORE and N-ACTION requests. But for some reason the Storage-SCP will later respond with an N-EVENT-REPORT with EVENT_ID == 2 indicating that persistent storage failed for this one image.
In this case the modality probably wants to resend the same image at a later point in time: should it send it with a complete new SOP Instance UID (obviously since the standard mandates unique UIDs) or reuse the old one in hope the SCP can resolve it (normally would lead to an 0x0111 failure at C-STORE level, wouldn't it?). Thought further in that case what about the Referenced Image Sequence within a MPPS N-SET request that mabye was also already successfully send by the modality, in the case described above it would include a dangling instance (the one that could not be stored by the storage SCP) and would miss the one instance that was resend by the modality at the later time.

What is your opinion on such a scenario described in general and how would you want a modality to behave in such a case?

Best regards,
Morty

Different Storage-SCP implementation, different view point:

"If an instance with the same SOP Instance UID is already present in the archive then the received instance will replace the old instance. So, if a remote node sends twice the same instance (sameSOP Instance UID) the old instance will be replaced by the latest received instance."

When you resend the same image you shall use the same SOP Instance UID and not create a new one.
The receiving station has indicated it didn't store it correctly as you said the Commitment failed. So will have no conflict when the image arrives again.
Your statement (obviously since the standard mandates unique UIDs) is in that way not correct as it is the same object content and shall have the same UID.
How systems handle when they receive the same object twice is up to the implementation, some will store the latest version others will ignore the latest version.
The quoted Conformance Statement gives you a clear indication what the system is doing.

Op vrijdag 18 februari 2022 om 16:19:28 UTC+1 schreef madMorty:
> Different Storage-SCP implementation, different view point:
>
> "If an instance with the same SOP Instance UID is already present in the archive then the received instance will replace the old instance. So, if a remote node sends twice the same instance (sameSOP Instance UID) the old instance will be replaced by the latest received instance."