ok, so you run this on target.... I guess something is strange there... I might try tomorrow if I catch time as I never used other two approaches (nor I plan to).

I usually always use first method... but, on which host did you execute 2 other which have failed?  Target host or some admin host?

Everything was repeated to show how weird the failure was. Same client, same server.

Does that answer your question regarding target?

Here I'm... got delayed by bizzare issue (running boinc 24/7 for past 6 months and keeping 8 cores at 100% resulted in sudden death of power supply... where is this world going to...).

Test 1 - what works

$HOST2 C:\>recover -s $SERVER -c $HOST1

Current working directory is C:\

recover> add boot

1 file(s) marked for recovery

recover> relocate C:\restore

recover> recover

Recovering 1 file from C:\ into C:\restore

Volumes needed (all near-line):

        TC200082 at TC2

Requesting 1 file(s), this may take a while...

Requesting 1 recover session(s) from server

C:\restore\Boot\

Received 1 file(s) from NSR server `$SERVER'

Recover completion time: 22-11-2011 15:51:12

recover> q

Test 2: Winblows GUI

That works too.

Test 3: CLI

$HOST2 C:\>recover -s $SERVER -R $HOST2 -c $HOST1 -iR -d C:\restore3 -a C:\Boot

Initiating remote recover to $HOST2 from server $SERVER, this may take a while...

Recovering 1 file from C:\ into C:\restore3

Requesting 1 file(s), this may take a while...

Requesting 1 recover session(s) from server

C:\restore3\Boot\

Received 1 file(s) from NSR server `$SERVER'

Recover completion time: 22-11-2011 16:03:37

Not sure why it fails in your case, but it looks as it had some sort of problems accessing defined port range.