checkpoint file unavailable: status.cpt (No such file or directory).
24 Jan 2011 16:35:36 UTC
Topic 195600
(moderation:
"checkpoint file unavailable: status.cpt (No such file or directory)."
New machine running Linux Ubuntu 10.10 amd64 (64bit).
I've surely managed to omit something 32bit
to get this error, but I am having trouble
figuring out what.
What have I failed to install?
task 216590480
workunit 91842090
I did just make another pass adding libraries and
whatever I could find 32bit. But I would
appreciate a hint about what I may have omitted.
Thanks in advance.
checkpoint file unavailable: status.cpt (No such file or directo
)
Not the boss of all BOINC then, as he'd know. ;-)
Try ldd boinc or ldd einstein_S5GC1HF_2.07_x86_64-pc-linux-gnu
That's el-dee-dee, and I am not completely sure that's the correct Einstein binary, please check in your BOINC\projects\einstein.phys.uwm.edu\ directory.
Did you get the 64bit application or the 32bit application?
I should have thought of ldd
)
I should have thought of ldd myself. Thanks for the reminder.
Was busy revamping the new system's cabling much of the day,
so I did not get back to addressing the real problem very quickly.
I had trouble finding the missing libraries detected by ldd on:
/var/lib/boinc-client/projects/einstein.phys.uwm.edu/einsteinbinary_BRP3_1.05_i686-pc-linux-gnu__BRP3SSE
the following was already ok.
/var/lib/boinc-client/projects/einstein.phys.uwm.edu/einsteinbinary_BRP3_1.00_graphics_i686-pc-linux-gnu
A bash script named 'getlibs' I found:
http://frozenfox.freehostia.com/cappy/getlibs
does all the heavy lifting and it found the right packages
for me (till all the dependencies were
satisfied). Highly recommended.
I found it at http://ubuntuforums.org/showthread.php?t=474790
I worry a bit that (because libdl.so is linked in) some
dynamically-requested shared-library I do not have will be needed.
But I'll try with what I have now. Once I know if things are better
(or not) I'll post here in this thread.
Again, thanks for the hint.
Ah. To answer your question,
)
Ah. To answer your question, both apps are 32bit. It was
the 32bit versions missing from /usr/lib32.
Results are being generated
)
Results are being generated now, seemingly
useful results. Checkpoint records
being found. Thanks again for the help.
You're welcome. Always glad
)
You're welcome. Always glad to give a nudge. :-)
I notice the same problem:
)
I notice the same problem: Workunit 92131757. Although 'status.cpt' is stored in the data section (in the slots 0,2,4) the file is noted as missing.
RE: I notice the same
)
I'm not sure why you think there is a 'problem' with the workunit you linked to. The two tasks in that quorum both completed successfully and have been marked as valid. If you look at the two hostIDs that crunched the tasks, one belongs to 'Nemo' (part of the 1560 core Nemo cluster) and one is 'anonymous' so I guess that one is yours. If you click on that taskID you can peruse the output returned to the server and there are no [ERROR] messages anywhere to be seen.
There are some [INFO] messages which do mention that there was no checkpoint file to be found, but this is entirely normal. Each BRP3 task is made up of 4 sub-tasks packaged together. As each sub-task starts crunching, there will be no checkpoint file to start with until some crunching has been completed and a checkpoint saved. This fact is simply being recorded for information purposes, along with the comment that crunching is "Starting from scratch ...". You should be able to find these [INFO] messages exactly 4 times in the task output, unless a task happens to be started, stopped and restarted so quickly that there wasn't time for an initial checkpoint to be written. In that case you could find more than four sets of such messages.
Cheers,
Gary.
Thank you Gary for explaining
)
Thank you Gary for explaining the message about the unavailable checkpoint file. If this is a normal behaviour of the application then it is ok for me.