I have accidentally corrupted my BOINC directory on a host, and now try to resume operation using a backup copy.
Then I found that the computer BOINC self-reset itself, and register as a new computer. The venue also reset itself to empty.

This is what happens (version 5.10.30):
1) The client_state.xml and client_state_prev.xml is set to the correct computer, on venue "Work".
2) BOINC starts.
3) Messages:


2011-12-14 22:55:07||Starting BOINC client version 5.10.30 for windows_intelx86
2011-12-14 22:55:07||log flags: task, file_xfer, sched_ops
2011-12-14 22:55:07||Libraries: libcurl/7.17.1 OpenSSL/0.9.8e zlib/1.2.3
2011-12-14 22:55:08||Processor: 1 GenuineIntel x86 Family 6 Model 8 Stepping 6 403MHz [x86 Family 6 Model 8 Stepping 6]
2011-12-14 22:55:08||Processor features: fpu tsc sse mmx
2011-12-14 22:55:08||OS: Microsoft Windows 2000: Professional Edition, Service Pack 4, (05.00.2195.00)
2011-12-14 22:55:08||Disk: 24.41 GB total, 17.58 GB free
2011-12-14 22:55:08||Local time is UTC +1 hours
2011-12-14 22:55:08|World Community Grid|URL: http://www.worldcommunitygrid.org/; Computer ID: *****; location: work; project prefs: work


So the BOINC starts on the correct venue "Work", with the correct ID from the backup.


2011-12-14 22:55:09||General prefs: from World Community Grid (last modified 11-Dec-2011 23:13:20)
2011-12-14 22:55:09||Host location: work
2011-12-14 22:55:09||General prefs: using separate prefs for work
2011-12-14 22:55:09||Reading preferences override file

2011-12-14 22:55:09||Running CPU benchmarks
2011-12-14 22:55:09||Suspending network activity - user request


Then the messages stop, until I re-enable network access.
Then the "Work" venue is lost, and so also is the computer ID.


2011-12-14 22:55:49||Resuming network activity
2011-12-14 22:55:49|World Community Grid|Started download of stat_v01.png
2011-12-14 22:55:49|World Community Grid|Started download of default_00_v01.gif
2011-12-14 22:55:50|World Community Grid|Sending scheduler request: To fetch work. Requesting 1 seconds of work, reporting 0 completed tasks
2011-12-14 22:55:55|World Community Grid|Scheduler request succeeded: got 0 new tasks
2011-12-14 22:55:55|World Community Grid|Message from server: No work sent

2011-12-14 22:55:55|World Community Grid|New host venue:
2011-12-14 22:55:56||General prefs: from World Community Grid (last modified 11-Dec-2011 23:13:20)
2011-12-14 22:55:56||Host location: none
2011-12-14 22:55:56||General prefs: using your defaults
2011-12-14 22:55:56||Reading preferences override file

2011-12-14 22:55:56|World Community Grid|Generated new host CPID: c0a2323be39f7bdecda677eff05cd238


The problem is that the server and the BOINC is out-of-sync on the message counter:
Data directory/client_state.xml <project>
<rpc_seqno>2514</rpc_seqno>

It could be difficult to guess the correct/actual <rpc_seqno>.
Simply lift the number to be greater than the one on the server -- increase by 3 per day, approx. When the number is larger than the server value, the server correctly identify the computer as your old computer!