I had also a bunch of errors from several machines. It seems to me that it happened when the current in-progress-WU finished and the cached WUs are the old version.

One machine (my slowest) finished a longer running HPF2 WU at 11:14 that validated OK. 3 cached HCC WUs errored out immediately with this error:
<core_client_version>6.2.28</core_client_version>
<![CDATA[
<message>
app_version download error: couldn't get input files:
<file_xfer_error>
<file_name>wcg_hcc1_img_6.08_windows_intelx86</file_name>
<error_code>-120</error_code>
<error_message>signature verification failed</error_message>
</file_xfer_error>

</message>
]]>

Cache was filled with new WUs (4 HCC), but I don't have access to that machine atm, and there are no wingman results yet to tell about the app version.

My faster machine (2 core) reported 3 WUs (1 HCC, 2 HPF2) at 03:19 which are valid (1 HPF2 and HCC) resp. PV (1 HPF2). All these old versions. At the same time the machine received 1 HCC 6.40 unit.
At 06:18 another HCC with old version was reported that is now in PV. At the same time 18 WUs (3 C4CW, 8 HPF2, 7 HCC) were dumped with errors. Cache was refilled at that time with 11 HPF2 and 3 HCC (all version 6.40). From the fresh downloads 2 HCC and 2 HPF2 are already in PV. (= finished without error)

One other machine (also a slow one) is still crunching on a long running HPF2 WU, and still has 1 HPF2, 1 HCC and 1 C4CW with old versions in cache. I will watch what happens when the current WU finishes. I expect that the cached units are then dumped and the cache refilled.

All machines (except the first one listed above) are running Boinc 6.10.58 on Windows 32-bit (XP for the first 2 listed above, Win7 for the last one)

HTH to understand what the reason is for this.

Greetings
Thorsten

I also lost several WUs in 5 different machines. 3 of them only run DDDT2 and the other 2 only run HCMD2. Most WU ended in error with 0 CPU Time, but some of them finished, for example:

CMD2_ 1544-1JWY_ A.clustersOccur-3BRW_ B.clustersOccur_ 38_ 91610_ 92351_ 0-- 615 Error 2/27/11 23:07:58 3/1/11 16:36:50 9.63 48.9 / 0.0

Log:

Result Name: CMD2_ 1544-1JWY_ A.clustersOccur-3BRW_ B.clustersOccur_ 38_ 91610_ 92351_ 0--
<core_client_version>6.2.28</core_client_version>
<![CDATA[
<stderr_txt>
INFO: No state to restore. Start from the beginning.
called boinc_finish

</stderr_txt>
<message>
app_version download error: couldn't get input files:
<file_xfer_error>
<file_name>wcg_hcmd2_maxdo_6.15_windows_intelx86</file_name>
<error_code>-120</error_code>
<error_message>signature verification failed</error_message>
</file_xfer_error>

</message>
]]>

Just to note that it happens with any non-zero cache, it's just less obvious. In the short cache scenario, the older cached WU fails and is replaced with the newer cached 6.40 WU in less than a second, so I'm guessing most users with a short cache haven't even noticed.

e.g.
01-Mar-2011 20:41:41 [World Community Grid] Starting CMD2_1551-2QZU_A.clustersOccur-2RC4_A.clustersOccur_7_1
01-Mar-2011 20:41:41 [World Community Grid] [error] Signature verification failed for wcg_hcmd2_maxdo_6.15_i686-pc-linux-gnu
01-Mar-2011 20:41:41 [World Community Grid] Starting CMD2_1549-2QZU_A.clustersOccur-2Q7N_A.clustersOccur_15_64357_65107_64618_65107_0
01-Mar-2011 20:41:41 [World Community Grid] Starting task CMD2_1549-2QZU_A.clustersOccur-2Q7N_A.clustersOccur_15_64357_65107_64618_65107_0 using hcmd2 version 640
01-Mar-2011 20:41:42 [World Community Grid] Computation for task CMD2_1551-2QZU_A.clustersOccur-2RC4_A.clustersOccur_7_1 finished
01-Mar-2011 20:41:42 [World Community Grid] Output file CMD2_1551-2QZU_A.clustersOccur-2RC4_A.clustersOccur_7_1_0 for task CMD2_1551-2QZU_A.clustersOccur-2RC4_A.clustersOccur_7_1 absent

If you check how many WUs have dumped with 0 elapsed time and error -120 or -163, I guess you'll find huge numbers.

I agree with KWSN - A Shrubbery. I have lots of machine running and none have this problem. Is this problem only with DDDT2? because I am not running that at this time. Is this going to be a continuing problem until all of the old version get cleared out? Or is it more of a time thing?

I run BOINC 6.10.56 on all my machines.

I just noticed my WUs for one machine were all terminating with Error and it's not restricted to DDDT2. For example, from Results Status, is the following:

Result Name Device Name Status Sent Time Time Due /
Return Time CPU Time (hours) Claimed/ Granted BOINC Credit
ts01_ b137_ pr02a0_ 1-- DAVETHOMPSON-PC Error 2/27/11 02:21:06 3/2/11 03:15:53 0.00 0.0 / 0.0
ts01_ b129_ pr45b1_ 1-- DAVETHOMPSON-PC Error 2/26/11 23:15:27 3/2/11 03:15:53 0.00 0.0 / 0.0
ts01_ b124_ pca009_ 0-- DAVETHOMPSON-PC Error 2/26/11 21:55:59 3/2/11 03:15:53 0.00 0.0 / 0.0

This is on four pages of the Results Status for a total of 60 WUs in Error. All of the WUs are for DDDT2 except one which is for CEP2. One of the first Error returns is the following:

Result Log

Result Name: ts01_ b096_ pr45b1_ 0--
<core_client_version>6.10.18</core_client_version>
<![CDATA[
<message>
app_version download error: couldn't get input files:
<file_xfer_error>
<file_name>wcg_dddt2_charmm_6.17_windows_intelx86</file_name>
<error_code>-120</error_code>
<error_message>signature verification error</error_message>
</file_xfer_error>

</message>
]]>

This does not appear for most of the other Errors. I exited BOINC on that one machine and restarted; things seem to be running normally.