Comment 2 for bug 622844

This is perhaps a duplicate of bug 532550. It appears that while sending the new revisions to the server (with an 'insert_stream' HPSS call) the server has seen a corruption in the stream from the client and aborted the stream with this error.

Like 532550 the cause is a mystery. Is there anything unusual about the branch being pushed? Is it very large? Are there non-ascii filenames in the new revisions? (I'd be surprised if committed filenames were the culprit, but that was a theory proposed for 532550, so worth checking).

And is this error reproducible?