"Next benchmark" is STUCK...

** Reply to message from WD Loughman <***@earthlink.net> on Sun, 30 Jan
2011 17:34:11 -0800

Post by WD Loughman
...and has been since 1-14-2011.
Hi all. I'm back, using BOINC5.

Me, too. Boinc6 is a horrific memory hog that I cannot afford to run on my old,
resource-constrained-compared-to-current-standards PC.

Post by WD Loughman
BUT, despite the length of time, and having received 56 WUs in one big
swoosh on that day, there has been no benchmarking at all since.

Are you logging Boinc messages to a file? If so, have you verified that there
have been no benchmarks in this period?

Post by WD Loughman
"Next benchmark :904 days -12:54:20" , each and every day since. No
change at all!
WUs are being crunched, uploaded and processed. I'm getting credit,
which gets posted.

So it is just the "Next benchmark" that is frozen? The rest of the JBSWU
display changes?

If you want help debugging the problem, ...
1) Please run the debug version of JBSWU. (Note: With the options I ask you to
use, the debug files created may be several MB in size. So make sure there is
that much free space on the "current drive" before starting this program.)
2) It will prompt you for a variety of options. Say "Yes" to the client state
dump and the two REXX trace options (and "No" to all the others)
3) After it starts, type "x" (for exit) as soon as the display shows the bogus
"Next Benchmark" time.
4) After JBSWU (Debug) exits it will prompt you about zipping up the debug
files (if ZIP.EXE is in the PATH). Say "yes" and email me directly (not thru
this list) with the zip file attached. (If ZIP.EXE is not on the PATH or if
you choose not to zip up the files, JBSWU (Debug) will display a list of
filename masks to use when you are ready to zip up the appropriate files.)

Post by WD Loughman
Several times I've put "0" in the correct "_rpc" spot in the
client_state.xml file - no effect. "Next benchmark" never changes,
*stays* at exactly what I wrote just above.
--> Any ideas, please?

Have you tried to start a benchmark manually? If not, you can use REXXRPC to do
this. (If you are going to run JBSWU (Debug) as described above, please wait
until that is done before you possibly "fix" the problem by running a
benchmark.) The syntax for starting a benchmark manually is:
REXXRPC --run_benchmarks

Post by WD Loughman
I want to shift over to BOINC6, try it again anyway, and I want to quit
BOINC5 gracefully. That is, do *all* the current supply (only 3 still
queued), and shut down *before* I get any more.
--> Please, what are the right boinc_monitor.cmd parameters to do that?

When I switched from Boinc6 back to Boinc5 I
1) Stopped the fetch of new WU's with:
rexxrpc --project setiathome.berkeley.edu nomorework
2) Waited until all queued WU's were finished and uploaded
3) Stopped Boinc5
4) Started Boinc6. (I have them installed in separate directories.)
5) I can't remember if I had to use:
rexxrpc --project setiathome.berkeley.edu allowmorework
to get more WU's.

--
John Small

------------------------------------

WD Loughman

2011-01-31 20:27:47 UTC

Post by John Small
2011 17:34:11 -0800

Post by WD Loughman
...and has been since 1-14-2011.
Hi all. I'm back, using BOINC5.

Me, too. Boinc6 is a horrific memory hog that I cannot afford to run on my old,

if called_as == 'COMMAND' then
do
say "Current network mode: "previous_network_mode
say "Current run mode : "previous_run_mode
end

if called_as == 'COMMAND' then
do
say "Current network mode: "previous_network_mode
say "Current run mode : "previous_run_mode
end

Damn. I'm still going to try it again. Others seem to manage, and
*now* I've a new and very much up-to-date machine.

Post by John Small
resource-constrained-compared-to-current-standards PC.

IMO it's hard-coded size of the Shared Memory Arena doing-in all of us.
A built-in OS/2 constraint which isn't going to change anytime soon.

Post by WD Loughman
BUT, despite the length of time, and having received 56 WUs in one big
swoosh on that day, there has been no benchmarking at all since.

I was wrong. Log shows it *is* happening - just not being shown by
JBSwuMon.

Post by John Small
Are you logging Boinc messages to a file? If so, have you verified that there
have been no benchmarks in this period?

I use stdoutdae.txt, both to monitor with LogView and to save
periodically as an ongoing 'audit trail'. It shows benchmarking *has*
occurred. Just, it's not reflected in JBSwuMon's display.

Post by WD Loughman
"Next benchmark :904 days -12:54:20" , each and every day since. No
change at all!
WUs are being crunched, uploaded and processed. I'm getting credit,
which gets posted.

So it is just the "Next benchmark" that is frozen? The rest of the JBSWU
display changes?

Just "Next benchmark". All other functions/display appear to be normal.

Post by John Small
If you want help debugging the problem, ...
1) Please run the debug version of JBSWU. (Note: With the options I ask you to
use, the debug files created may be several MB in size. So make sure there is
that much free space on the "current drive" before starting this program.)
2) It will prompt you for a variety of options. Say "Yes" to the client state
dump and the two REXX trace options (and "No" to all the others)
3) After it starts, type "x" (for exit) as soon as the display shows the bogus
"Next Benchmark" time.
4) After JBSWU (Debug) exits it will prompt you about zipping up the debug
files (if ZIP.EXE is in the PATH). Say "yes" and email me directly (not thru
this list) with the zip file attached. (If ZIP.EXE is not on the PATH or if
you choose not to zip up the files, JBSWU (Debug) will display a list of
filename masks to use when you are ready to zip up the appropriate files.)

I'll do that later today. Needful chores ( "Yes, dear!" ) come first. ;)

Have you tried to start a benchmark manually? If not, you can use REXXRPC to do

Yes. Both as 'Benchmark.cmd' and "B" withing JBSwuMon.

Benchmark.cmd "opens" an "invisible" window with contains:
--------------------------------------
Current network mode: previous_network_mode
Current run mode : previous_run_mode
--------------------------------------
... if my memory serves. But it never closes. I can close it manually
if I choose, but it seems harmless and seems never to *do* anything.

Post by John Small
this. (If you are going to run JBSWU (Debug) as described above, please wait
until that is done before you possibly "fix" the problem by running a
REXXRPC --run_benchmarks

Haven't done that yet. ...Later

Too late. I've received and am crunching another 20 WUs.

Post by John Small
When I switched from Boinc6 back to Boinc5 I
rexxrpc --project setiathome.berkeley.edu nomorework
2) Waited until all queued WU's were finished and uploaded
3) Stopped Boinc5
4) Started Boinc6. (I have them installed in separate directories.)
rexxrpc --project setiathome.berkeley.edu allowmorework
to get more WU's.

Thanks John, for (all) the above.
I'll get to all of it later, after Herself is mollified. And after I
transfer/clone all my HDDs to new ones. There was some damage after
all, which I fixed. And/but now I've got to move everything (500-or-so
GB) to safety.

- Bill

--
WD "Bill" Loughman - Berkeley, California USA
http://home.earthlink.net/~wdloughman/wdl.htm

------------------------------------

John Small

2011-01-31 21:17:10 UTC

** Reply to message from WD Loughman <***@earthlink.net> on Mon, 31 Jan
2011 12:27:47 -0800

Post by John Small
2011 17:34:11 -0800
Have you tried to start a benchmark manually? If not, you can use REXXRPC to do

Yes. Both as 'Benchmark.cmd' and "B" withing JBSwuMon.
--------------------------------------
Current network mode: previous_network_mode
Current run mode : previous_run_mode
--------------------------------------
... if my memory serves. But it never closes. I can close it manually
if I choose, but it seems harmless and seems never to *do* anything.

Your trouble with benchmark.cmd caused me to look at the code again. I can see
how it might fail. What is the current/working directory when you run
benchmark.cmd ? If the current/working directory is not the directory where
REXXRPC is located and REXXRPC is not on the PATH, then I could see how it
would not work. My next release will fix this.

If this is not the cause of your problem, then it's possible that REXXRPC is
broken somehow. This would explain both the problem with benchmark.cmd and with
the frozen "Next Benchmark" time in JBSWU.

You said earlier that only the "Next Benchmark" time was frozen. But you only
mentioned that the credit data gets updated. Does the other, non-credit data
get updated in JBSWU? The data in the last three lines of the JBSWU display
should change after every cycle. The default cycle length is 30 seconds. If the
data at the bottom is not being updated either, then this points to a broken
REXXRPC. I suggest that you try reinstalling REXXRPC and see if that fixes both
problems. (Try this before you try the other debug procedure I gave you
earlier.)

--
John Small

------------------------------------

WD Loughman

2011-02-02 09:18:18 UTC

Post by John Small
2011 12:27:47 -0800

Post by John Small
2011 17:34:11 -0800
Have you tried to start a benchmark manually? If not, you can use REXXRPC to do

Memory did not serve; not well:

[e:\boinc5\bin]benchmark.cmd
Current network mode: NETWORK_MODE
Current run mode : RUN_MODE

Post by WD Loughman
But it never closes. I can close it manually
if I choose, but it seems harmless and seems never to *do* anything.

ALL your commands related to BOINC5 are in "e:\Boinc5\bin".
Themselves: neither BOINC nor SETI are defined in any PATH.

Post by John Small
If this is not the cause of your problem, then it's possible that REXXRPC is
broken somehow. This would explain both the problem with benchmark.cmd and with
the frozen "Next Benchmark" time in JBSWU.

Your call...

Post by John Small
You said earlier that only the "Next Benchmark" time was frozen. But you only
mentioned that the credit data gets updated. Does the other, non-credit data
get updated in JBSWU? The data in the last three lines of the JBSWU display
should change after every cycle. The default cycle length is 30 seconds. If the
data at the bottom is not being updated either, then this points to a broken
REXXRPC.

ALL the lines are updated (when appropriate); everything seems "normal"...
EXCEPT the "Next benchmark" line remains unchanged.

Post by John Small
I suggest that you try reinstalling REXXRPC and see if that fixes both
problems. (Try this before you try the other debug procedure I gave you
earlier.)

For the hell of it, I did that - from what I've archived as
"rexxrpc251.zip" .
That is: 9-13-10 15:13 25,108 154 rexxrpc251.zip

Before re-install: 9-13-10 8:10 33,947 32,767 rexxrpc.cmd
After re-install: 9-13-10 8:10 33,947 32,767 rexxrpc.cmd

...no change in behavior. "Next benchmark" still stuck.
I'll wait on anything else, till after you play with the rexxrpc code.

- Bill

--
WD "Bill" Loughman - Berkeley, California USA
http://home.earthlink.net/~wdloughman/wdl.htm

------------------------------------

John Small

2011-02-02 12:14:58 UTC

** Reply to message from WD Loughman <***@earthlink.net> on Wed, 02 Feb
2011 01:18:18 -0800

First, it seems you have two problems:
1) The "Next benchmark" time is frozen in JBSWU
2) BENCHMARK.CMD does not seem to be working for you.
These may be separate problems or they may be related in some way.

Post by John Small
2011 12:27:47 -0800

Post by John Small
2011 17:34:11 -0800
Have you tried to start a benchmark manually? If not, you can use REXXRPC to do

[e:\boinc5\bin]benchmark.cmd
Current network mode: NETWORK_MODE
Current run mode : RUN_MODE

Post by WD Loughman
But it never closes.

Problem 2: Both the bogus modes and the fact that it never suggests to me that
the REXXRPC being used by BENCHMARK.CMD is not working.

Post by WD Loughman
... I can close it manually
if I choose, but it seems harmless and seems never to *do* anything.

This is because all calls to REXXRPC are failing. BENCHMARK.CMD is waiting
(forever) for a benchmark that was never started to finish!?

Post by John Small
Your trouble with benchmark.cmd caused me to look at the code again. I can see
how it might fail. What is the current/working directory when you run
benchmark.cmd ? If the current/working directory is not the directory where
REXXRPC is located and REXXRPC is not on the PATH, then I could see how it
would not work. My next release will fix this.

ALL your commands related to BOINC5 are in "e:\Boinc5\bin".
Themselves: neither BOINC nor SETI are defined in any PATH.

So it is not in the PATH. But how do you run BENCHMARK.CMD?
-- Clicking on some icon? If so, what is the current/working directory for this
icon?
-- From the command line? And if so, what is the current directory when you run
it?
-- Other?

Your call...

ALL the lines are updated (when appropriate); everything seems "normal"...
EXCEPT the "Next benchmark" line remains unchanged.

"When appropriate"? The "CPU Time" and "Time to go" should change every time
the display is updated (which by default is every 30 seconds).