I'm trying to configure Rosetta 2.05 to behave under the auspices of BOINC v6.10.18 intalled on a Win2003 R2 (Standard) primary-desktop system. It is necessary to establish necessary minimum sandbox paramters for Rosetta to function within the purvue of Comodo - v3.14.130099.587 - Internet Security firewall. Thus far I have established the following configuration:

BOINCmngr.exe has priveleges to execute:

boinc.exe
boinc.scr
boincscr.exe
iexplore.exe
%install_Data%projects**

in addition to DNSRPC client service priveleges, it has explicit permissions to access and modify the following registry keys:

HKLMSYSTEMControlSet001ServicesEventlogApplicationESENTEventMessageFile

HKLMSYSTEMControlSet001ServicesEventlogApplicationESENTCategoryMessageFile

HKLMSYSTEMControlSet001ServicesEventlogApplicationESENTCategoryMessageFile

HKLMSYSTEMControlSet001ServicesEventlogApplicationESENTCategoryCount

HKLMSYSTEMControlSet001ServicesEventlogApplicationESENTTypesSupported

BOINC.exe execute priveleges are restricted to:

%install_Data%projects**

read/write priveleges are:

%install_Data%*
%install_Data%projects*
%install_Data%slots*

In addition to DNS ClientRPC Client service privelege, BOINC.exe has the same priveleges to access and modify the aforementioned registry keys.

BOINC.exe, BOINCmngr.exe, BOINC.scr & BOINCscr.exe have the following network access priveleges:

TCP out from 0.0.0.0 to 127.0.0.1 (various port ranges)

NIC to DNS (port 53 only)
NIC to BOINC project hosts (port 80)
NIC to SETI_DataServer IP (port 80)
NIC to Rosetta_DataServer IP (port 80)
NIC to Einstein_DL IP (port 80)
NIC to Lattice IP (port 80)
NIC to BOINC_net (1e100.net) (port 80).

Despite all of that, the following message erupted:

3/11/2010 4:53:01 AM rosetta@home Output file quota_6_1vcc__BOINC_abrelax.v1_SAVE_ALL_OUT_18319_1101_0_0 for task quota_6_1vcc__BOINC_abrelax.v1_SAVE_ALL_OUT_18319_1101_0 absent

---

FYI:

Rosetta started at 03:41 (EST) and despite the aformentioned configuration aborted with the aforementioned message at 04:53 (EST).

Rosetta has NO problem downloading any WU, nor connecting to the project web-site, etc.

Nor does Rosetta have any problem initiating processing on any arbitrary WU. It appears that "work" is actually being done, i.e., CPU useage @ 100% and CPU heat goes up.

---

I've just suspended ALL projects except Rosetta.

Moreover, I just aborted ALL the WU that just came down except the ONE that's crunching (my bad - I left a hole) .

That's ALL that's going on here: Rosetta 2.05

If this can not be expeditiously be resolved: I'll detatch.

---

3/11/2010 5:16:56 AM rosetta@home Computation for task quota_6_1eyvA_BOINC_abrelax.v1_SAVE_ALL_OUT_18319_1118_0 finished
3/11/2010 5:16:56 AM rosetta@home Output file quota_6_1eyvA_BOINC_abrelax.v1_SAVE_ALL_OUT_18319_1118_0_0 for task quota_6_1eyvA_BOINC_abrelax.v1_SAVE_ALL_OUT_18319_1118_0 absent

---

ALL new WU comeing now...

---

Running...

---

More WU being downloaded (still running)...

4 WU downloaded and 1 processing w/o single hicc-up (last D/L succeeded @ 05:27 (EST).

Now I restart Win2003. If there's any problems with permissions pertaining to anything BOINC, it'll be readily become manifest.

---

3/11/2010 6:13:41 AM rosetta@home Restarting task MC14A_boinc_loopbuild_threading_cst_relax_loopbuild_threading_cst_relax_firas_IGNORE_THE_REST_18628_4352_0 using minirosetta version 205
3/11/2010 7:00:13 AM rosetta@home Computation for task MC14A_boinc_loopbuild_threading_cst_relax_loopbuild_threading_cst_relax_firas_IGNORE_THE_REST_18628_4352_0 finished
3/11/2010 7:00:13 AM rosetta@home Output file MC14A_boinc_loopbuild_threading_cst_relax_loopbuild_threading_cst_relax_firas_IGNORE_THE_REST_18628_4352_0_0 for task MC14A_boinc_loopbuild_threading_cst_relax_loopbuild_threading_cst_relax_firas_IGNORE_THE_REST_18628_4352_0 absent
3/11/2010 7:00:13 AM rosetta@home Starting lr5_dun08_slow_corrections_semirot_it02_A_rlbd_1unp_SAVE_ALL_OUT_IGNORE_THE_REST_DECOY_18081_971_0
3/11/2010 7:00:17 AM rosetta@home Starting task lr5_dun08_slow_corrections_semirot_it02_A_rlbd_1unp_SAVE_ALL_OUT_IGNORE_THE_REST_DECOY_18081_971_0 using minirosetta version 205


I've detached from this project; Rosetta has anything it could possibly need (I do not understand what the hold up is).

---

Reattached.

Allow: no new tasks.

Aborted downloaded task.

Resource share = 1

---

Access Violation (0xc0000005)

Rosetta is quite memory-hungry - my rosetta processes are using anything between 237-512MB RAM each at the moment so your 384MB will fill up quickly.

You might have some luck with the following settings in preferences:

Uncheck: Do work: while computer is in use
Memory - Use at most: 100% when computer is in use
Check: Leave applications in memory while suspended

HTH
Danny

---

Uh, what?

---

Uh, what?

Your tasks are exiting with this error: - exit code -1073741819 (0xc0000005)
E.g. this is one of your failed tasks: https://boinc.bakerlab.org/rosetta/result.php?resultid=323440599


That's a memory access violation. They're trying to access memory they shouldn't - I presume because they need more RAM than your BOINC preferences allow.

Try using the settings I gave previously in BOINC Manager...

---

What you said made absolutely no sense whatsoever; I was under the serious delusion that there were 3x256MB on-board. However, Win task manager distinctly concurred with you that I had some weird number about 384MB.

What? How's zat possible with 3x256? So I rebooted and took a look in BIOS and that was reaffirmed there too (despite 3 DIMMS being indicated full). What?

Upon closer inspection, it is discovered that the sticks are 128MB / ea.

So now I'm back to the original 3x256 ECC CL2 PC133 SDRAM I originally had before swapping out all the sticks for what I mistakenly thought were 256MB CL2 non-ECC SDRAM. All this transpired because of a mobo failure I had recently.

Worst that can happen now is that the NMI machine check memory parity errors STOP blue-screens begin to reoccure (which precipitated the memory swap to begin with). I seriously was delusional in my belief I had 3x256MB.

Thanx for the head-sup. Even so I DO have 2x512 non-ECC CL2 PC133 SDRAM on the way, I REALLY thought I had 3x256MB. Man this SUCKS.

---

Resource share = 100

---

What you said made absolutely no sense whatsoever; I was under the serious delusion that there were 3x256MB on-board. However, Win task manager distinctly concurred with you that I had some weird number about 384MB.

What? How's zat possible with 3x256? So I rebooted and took a look in BIOS and that was reaffirmed there too (despite 3 DIMMS being indicated full). What?

Upon closer inspection, it is discovered that the sticks are 128MB / ea.

So now I'm back to the original 3x256 ECC CL2 PC133 SDRAM I originally had before swapping out all the sticks for what I mistakenly thought were 256MB CL2 non-ECC SDRAM. All this transpired because of a mobo failure I had recently.

Worst that can happen now is that the NMI machine check memory parity errors STOP blue-screens begin to reoccure (which precipitated the memory swap to begin with). I seriously was delusional in my belief I had 3x256MB.

Thanx for the head-sup. Even so I DO have 2x512 non-ECC CL2 PC133 SDRAM on the way, I REALLY thought I had 3x256MB. Man this SUCKS.

ah- are they high density chips on a bx or some other incompatible chipset?

---

Nope, low density chips. The problem is I thought I had 3x256MB when the DIMMS I swapped in were actually 3x128. I'm virtually certain that's the issue here. While other projects have run just fine for the last two weeks, obviously Rosetta needs more physical RAM (despite having 1152MB swap file).

The ECC RAM I swapped out I've been using for years. However, because of repeated NMI machine check memory parity errors after replacing a fried mobo, I figured the mobo damaged one (or more) of the DIMM modules. I know it damaged the slot-1 P3 1000; CPU running at 70+ Deg. C. plus all sorts of other errors too (non NMI machine check related). So I swapped out the ECC RAM with non-ECC RAM I had on hand (stuff I bought before I got the ECC RAM); they're Crucial CL2 non-ECC PC133 DIMMS. I originally paid $175 / stick. I got the ECC SDRAM a couple years later for a mere $75/256MB stick.) To my chagrin, I forgot the non-ECC sticks were only 128MB sticks.

So I swapped the original 3 sticks of 256MB ECC SDRAM back in. Wouldn't you know it? Not a single NMI machine check error since this morning. Go figure. And I was getting them errors within minutes of each other. Machine was unuseable until I swapped all the RAM out with those 128MB sticks. Now I'm running on the original RAM with no errors.

Dunno. Memtest86 checked out on all the RAM too, so I don't know. Its a used mobo, maybe the mobo's bad. Its a used CPU too, and the 866 is screaming at 976, so dunno. I've got a P3-S 1.4Ghz coming, along with 2x512 MB non-ECC CL2 PC133 SDRAM. Should be here any day now.

---