Customer has reported that large packages, in this particular case Images files,are not finishing downloading. Those keep retrying from the beginning if the download fails in any giving point rather than keep going from the last checkpoint.
They have multiple remote Sites where the package servers on those Sites relies on a "master" package server for the SMP to get these packages. No hierarchy. The download is slow due to the bandwidth speed on those remote locations. They usually throttle during business hours.
We have noticed that these packages usually gets above 50% and then starts downloading again. The customer uses primarily UNC transfer for those packages.
At this point the major concern from the customer is not that UNC transfer may be aborted by some sort of process but that the packages start downloading from the beginning again rather than from the last valid point.
Looking at the package servers affected, the agent logs shows:
Known issue. UNC to UNC resume functionality is not working. The CreationDate was used as a LastWrite date, thus big packages which has those dates different were not resumed in some situations. This was visible when small package was coming to the queue and take the priority in downloading, preempting the big one from current download.
This issue has been fixed with our ITMS 7.6 release.
A pointfix is available for those in ITMS 7.5 SP1 HF3 or HF5. See attached "Pointfix_eTrack3723963_7_5_SP1_HF3.zip" and "Pointfix_eTrack3791270_SMA_7_5_SP1_HF5.zip"
Read Me information is available in the attachment itself as well.