climateprediction.net home page
Posts by Dave Jackson

Posts by Dave Jackson

1) Questions and Answers : Getting started : removal issue (Message 65468)
Posted 4 days ago by Profile Dave Jackson
Post:
Removing it, might be possible by going to your data directory and deleting account_climateprediction.xml

If that doesn't work and you can't get an answer here, I am pretty sure you would get one at the boinc forums
2) Message boards : Number crunching : New work Discussion (Message 65464)
Posted 8 days ago by Profile Dave Jackson
Post:
Don't bother with nested virt, it has a substantial impact on performance. I got it working on cloud machines with nested virt, but performance wasn't any better than an older Intel box.
I guessed there would be a substantial performance hit. If I do it, it will be more about proof of concept than anything. Certainly not going to be doing it regularly.
3) Message boards : Number crunching : New work Discussion (Message 65455)
Posted 9 days ago by Profile Dave Jackson
Post:
Dave, do you also have an Intel PC? If so, have you tried setting up a macOS VM on it? For me, I couldn't get it to work on Ryzen 5900X but did on i7-4790 (both Windows 10).


I have a Ryzen7. I have no problems running another instance of Ubuntu in VB or KVM but haven't managed to get the MacOS VM to work on it. At some point, I will try a clean install and see if I can get it going on that. I haven't yet tried nesting via the Ubuntu VM.
4) Message boards : Number crunching : New work Discussion (Message 65451)
Posted 9 days ago by Profile Dave Jackson
Post:
Yes, I had quite a few completed from that batch, then a bunch of them failed. No obvious reason why some worked and some didn't but the first eight or nine did complete.
5) Message boards : Number crunching : New work Discussion (Message 65449)
Posted 10 days ago by Profile Dave Jackson
Post:
More HADCM3S in testing. No use to me as I haven't managed to get the virtualisation working. And of course no guarantees about when/if these will translate into main site work, though I think it likely they will at some point.
6) Message boards : Number crunching : Windows Work Units (Message 65447)
Posted 13 days ago by Profile Dave Jackson
Post:
No Windows WU in the pipeline that anyone knows of, or has even heard rumors of.


Yes, both Mac and N216 Linux have had testing batches running last week that completed OK. I am assuming these will lead to main site work but when is anybody's guess.
7) Message boards : Number crunching : New work Discussion (Message 65433)
Posted 15 days ago by Profile Dave Jackson
Post:
The only project that works for my Linux machine is universe@home and even that one had about 3 days of uploads that were not going up until now. Even now, that is pretty spotty.


TN-Grid is working for me. Prime Grid also works but is of little interest to me. Prior to WCG going off air to move everything, Africa Rainfall Project was my go to when nothing available here. Despite the fact that I have run Ubuntu in a VM using virtual box on here, the projects that use a vm don't seem to be working for me.
8) Message boards : Number crunching : New work Discussion (Message 65431)
Posted 15 days ago by Profile Dave Jackson
Post:
Last batch of N216 testing tasks were successful and thetesting of Mac HADCM3s tasks seems to be good so hoping that there will be some main site work soon as nothing else is appearing on the test site, WCG is still down so I am restricted to third and fourth choice projects at the moment.
9) Message boards : Number crunching : New work Discussion (Message 65430)
Posted 18 days ago by Profile Dave Jackson
Post:
I have learned a bit more. At this stage, it is a proof of concept thing only so quite a long time before or if anything even makes it to the testing crew outside of people at Oxford. I probably got over-excited at seeing an entry on the server status page for the testing site.
10) Message boards : Number crunching : New work Discussion (Message 65427)
Posted 18 days ago by Profile Dave Jackson
Post:
Doubtless some way off assuming it does happen eventually but there are tasks on the testing site for ARM processors. - I shall have to get myself a PI. I see any broadening out of the potential user base as a good thing. Whether anything will happen to get more native work for windows is anyone's guess though much change there in the short to medium term looks unlikely.

Edit: In fact it is for a particular embedded chip and so won't run on a Pi without virtualisation so probably not going to be relevant to the main site.
11) Questions and Answers : Unix/Linux : Running 32-bit MacOS Tasks on Linux with KVM (Message 65425)
Posted 22 days ago by Profile Dave Jackson
Post:
I know about the workaround. I posted the sed line. :p


Oops. That will teach me to post when half asleep.
12) Questions and Answers : Unix/Linux : Running 32-bit MacOS Tasks on Linux with KVM (Message 65419)
Posted 23 days ago by Profile Dave Jackson
Post:
... though it seems that the file downloader is a bit overwhelmed at the moment, I keep having downloads fail and go to project backoff state.


See the discussion in New Work Discussion in the number crunching section for a workaround for the download problems. The best guess currently is that this is some change by Oxford Uni to their network which they neglected to tell the project about.
13) Message boards : Number crunching : New work Discussion (Message 65417)
Posted 23 days ago by Profile Dave Jackson
Post:
This worked for me to get a small batch of tasks (on Mojave VM on VBox on Windows10). However it seems to only work once, so to download a second batch one has to go through the process again. client_state.xml file must refresh itself periodically. I had to start a new set of tasks as suspended state prevents new task acquisition so can't try again for about a week. Hopefully this gets fixed by the project by then.
It needs to be done again because the lines being edited are for the files currently stuck. New lines that need editing will be added for the relevant files when more work is downloaded. I would have too check to see if the lines are deleted when the downloads complete or if it happens when the task is reported.
14) Message boards : Number crunching : New work Discussion (Message 65414)
Posted 23 days ago by Profile Dave Jackson
Post:
Well, for now, if you want to get something running, shut down your BOINC client (closing BOINC Manager should do it), head over to /Library/Application Support/BOINC Data, and something like this ought to help you (back up client_state.xml first).

sed -i -e s/'https:\/\/download.cpdn'/'http:\/\/download.cpdn'/ client_state.xml



Neat. If we get the same issue on Linux tasks I will remember that.

Thank you hopefully the more savvy of our Mack users can use this while waiting for a fix from the project.
15) Message boards : Number crunching : New work Discussion (Message 65408)
Posted 24 days ago by Profile Dave Jackson
Post:
I will send another email. :(

Edit: done shortly after I posted, Richard and another moderator have also emailed Andy.
16) Message boards : Number crunching : New work Discussion (Message 65403)
Posted 24 days ago by Profile Dave Jackson
Post:
I have sent an email to Andy.

Edit: Andy says no issues with any of the servers this morning and thinks it may just be the large number of people contacting the servers for work.

If the problem persists do post again ideally with http debug flag set (The one Richard suggested.)
17) Message boards : Number crunching : New work Discussion (Message 65397)
Posted 25 days ago by Profile Dave Jackson
Post:
Both of my MacVMs I just lit up are unable to download any files related to the tasks - they try (with valid internet connections and project connectivity - they've both been able to request tasks), but then end up with nothing transferred, retry a few times, and eventually end up with all the downloads in the "project pending" state for a few hours of backoff.


Can you enable in http debug in event log options, click on retry on one of the downloads and see what the output in the event log is? And disable the http debug afterwards unless you want to fill the event log with its output. - probably worth having that before sending an email to the project.
18) Message boards : Number crunching : New work Discussion (Message 65387)
Posted 28 Apr 2022 by Profile Dave Jackson
Post:
Eehhh... if you're getting segfaults on Intel and not AMD, or AMD and not Intel, then something is either fouled up with the hardware, or the binary is seriously screwed up.

Despite being in competition, they both try to make very sure that the binary behavior of programs is identical between then. When it's not, it's led to some interesting edge cases and security violations, but that's at application/OS interface stuff, not just compute behavior.


The differences between processor types and also between operating systems when most tasks for CPDN used to have tasks that would run on Windows, Mac or Linux are dealt with by the statistical program that analyses batch results rather than just individual task results. There are minor differences in the rounding algorithms used between different chips which is what leads to some tasks working on one architecture or one chip and not another. I may be wrong but I think that failures on one architecture then succeeding on another is less common than it used to be but it certainly still happens.
19) Message boards : Number crunching : New work Discussion (Message 65385)
Posted 27 Apr 2022 by Profile Dave Jackson
Post:
Nothing from testing that looks to be producing main site work in a hurry but at least I got a retread yesterday.

Edit: A seg fault rather than missing libraries but I can hope that running it on an AMD processor when it failed on an Intel machine might mean it will work!
20) Message boards : Cafe CPDN : Weather forecasting (Message 65382)
Posted 25 Apr 2022 by Profile Dave Jackson
Post:
Some years ago when I studied statistics, the most accurate way of forecasting the next day's weather was to say, "It will be the same as today"

How things have changed!


Next 20

©2022 climateprediction.net