nebulous replication left an empty file behind

[bills]

When updating a chip imfile today I ran into a case where it failed because the burntool table had zero size.

This is strange since the file existed at one time since the chip processing had completed successfully once.

Investigating further we found that there were two instances. One of which had zero size.

This seems to indicate that the replication process failed and left a file behind.

For reference the file was

neb://ipp042.0/gpc1/20100118/o5214g0235o/o5214g0235o.ota22.burn.tbl

I have fixed the zero sized file by copying the good one on top of it.

[watersc1]

Replication creates the empty file first at the server level, and leaves it to the client to fill. Client.pm:241 does the copy, and the return status must not always be correctly reflecting the failure, as there is no entry for the failure of that file in the replicate log.