climateprediction.net home page
Posts by jrapdx

Posts by jrapdx

1) Message boards : Cafe CPDN : MicroSoft and Windows (Message 54748)
Posted 4 Sep 2016 by jrapdx
Post:
I am running Linux on this machine, but I have one with Win10, and another has Win8.1. Haven't yet seen the September "update", but I'll be on the lookout for it. Thanks for the warning.

I don't know if it's possible to block updates on Win10. There are utilities that can stop some traffic with MS, but haven't yet tried to stop the update "service". Not sure if doing that would have bad side-effects. I liked it better when it was possible to review updates before installing them.

Looks like it may get worse before MS will have to rethink these update policies, especially for sophisticated users who can't afford to have work interrupted and the instability going on.
2) Message boards : Number crunching : Whats up with credits 2.4million secs crunching for zero? (Message 54483)
Posted 11 Jul 2016 by jrapdx
Post:
Looks like it's a widespread problem, certainly true for my computers, and I saw posts by others with the same issues. Hope it gets fixed pretty soon, logs indicate zips are being uploaded, so appears it's the accounting for the work that's fallen down. Unless of course, there's something I don't know, always a risk of that...
3) Message boards : Number crunching : No trickles on webpage (Message 54482)
Posted 11 Jul 2016 by jrapdx
Post:
Yup, similar problems here. It's been going on for a while, like a week or two, but it will get straightened out eventually, it always has in the past. (In the last year or so anyway.) The BOINC logs on my computers show the work is being uploaded, which is the important part of volunteering for the project's research. I'm glad it's not my job to maintain the project servers, seems a bear to keep all the balls in the air all the time.
4) Message boards : Number crunching : No trickles on webpage (Message 54459)
Posted 9 Jul 2016 by jrapdx
Post:
Looks like the same problem here: BOINC log shows zips uploaded but trickles absent on web. However the "Your computers" page shows "Last contact" within the last 2 hours. I guess there's a different function or server involved with "contact" vs. trickle uploads.

Any estimate when things will get back to "normal"?
5) Message boards : Number crunching : ANOTHER UPLOAD PROBLEM (Message 54405)
Posted 29 Jun 2016 by jrapdx
Post:
Similar issues here. Log says trickles uploaded, but not they're not showing in the CPDN tasks page for the computer. Don't know if this will be fixed when server software is upgraded later in the week, or if a separate issue altogether.
6) Message boards : Number crunching : wah2_eu25 work units are BIG (Message 54281)
Posted 10 Jun 2016 by jrapdx
Post:
Re: Windows 10. Yes it is a quagmire for the privacy-minded. There are a number of guides on the web re: how to configure to decrease exposure to MS snooping.

Also, some tools are available to help reduce MS tracking on Win10. This one seems to work and is fairly easy to use: see DisableWinTracking at
https://github.com/10se1ucgo/DisableWinTracking
7) Message boards : Number crunching : ANOTHER UPLOAD PROBLEM (Message 54175)
Posted 23 May 2016 by jrapdx
Post:
Getting the same error messages here, at least on this machine. However my other computers aren't reporting errors probably because they're running different WUs.

AFAICT the current upload failure is affecting hadam3p_eu/pnw, but not wah2_eu/pnw (which uploaded OK). IIRC it's been the other way around before when it was the wah2 WUs that were not getting through. I guess it depends on exactly which server(s) are malfunctioning.

Lately the project seems to be having a hard time keeping all its plates spinning simultaneously. Let's hope these problems are resolved soon!
8) Message boards : Number crunching : Total Credit (Message 53980)
Posted 18 Apr 2016 by jrapdx
Post:
In case it helps, same problems here: no connection for uploading wah2 zips, starting 2016-04-17 06:19:06 UTC according to the event log. Also, though less important, credits haven't changed since Apr 10. Hopefully the outages will be fixed soon.
9) Message boards : Number crunching : Total Credit (Message 53952)
Posted 11 Apr 2016 by jrapdx
Post:
A nice surprise! I checked my "main" computer and unexpectedly there was a big jump in "Work done". I'll keep a good thought CPDN is back up and stays that way.

Thanks to you and Andy for get it going again!!
10) Message boards : Number crunching : Total Credit (Message 53946)
Posted 9 Apr 2016 by jrapdx
Post:
Maybe it's a different issue, but my credits haven't changed at all in 5 days, though as before tasks are still progressing and completing. Leads me to think there continue to be problems with CPDN's network or servers. Making it all work is a job I'm glad not to have, seems to be difficult to get all the plates spinning at the same time.
11) Message boards : Number crunching : Total Credit (Message 53937)
Posted 8 Apr 2016 by jrapdx
Post:
As I understand it trickles were introduced to compensate for the length of some of the earlier climate models that could take months to complete.
Nowadays the time is much shorter. Would it offload the servers if trickles were eliminated as credit sources,and credit issued on the completion of a work unit?

Trickles provide partial credit for work done, and I think eliminating this credit could provoke strong negative reactions. IIRC the duration of tasks is not the only issue that trickles address but also the problem that units are subject to crashing prior to completion sometimes after running for several weeks. IOW such occurrence is in effect random and unpredictable. So obviously there would be a huge risk of uproar from the community if credits were awarded only if and when tasks finished without errors.
12) Message boards : Number crunching : Total Credit (Message 53931)
Posted 7 Apr 2016 by jrapdx
Post:
Probably that explains why credits haven't changed (again) for the last few days. I can see the trickles are being uploaded, so the totals will eventually reflect the work being done. We'll see what happens Saturday!
13) Message boards : Number crunching : Total Credit (Message 53915)
Posted 7 Apr 2016 by jrapdx
Post:
That's odd about your computers. My pure Linux machines, which underwent Wine surgery in January, are only listed as being older than 30 days.

Certainly seems odd to me too. Prior to the crash the Linux OS had indeed faded into the older than 30 days list. I had assumed the reappearance of the Linux side was the result of "reading old tapes" when things were messed up, but don't know what's going on now.

Curiously during the outage Wine BOINC/CPDN was running flawlessly, but today the system "froze" again requiring manual reboot. Didn't cause any tasks to fail, so that's a good thing. Still seems like something's flaky in this system but not yet been able to pin it down.
14) Message boards : Number crunching : Total Credit (Message 53905)
Posted 6 Apr 2016 by jrapdx
Post:
Credits have increased a good bit, and recent avg too. Of course hard to know if the figures are "correct", but at least in the right direction and that's encouraging.

One thing is still puzzling. The "Your Computers" page shows my Linux machine twice, one listing is is for Linux OS, the other Windows. This makes sense insofar as it is a Linux box running BOINC under Wine. However the only BOINC client running is the Windows/Wine version. I haven't had the Linux BOINC client online since December 2015. It's still showing avg credit of 209.94, which is slightly lower than yesterday's reported average.

I'd attributed this misrepresentation to the recent/current server crash and maybe that's still affecting the report. It leaves a question whether the other credits are accurate. BTW the Linux/Windows "computers" have separate IDs: Windows == 1389186, Linux == 1373243. Should add that prior to the crash the Linux side wasn't showing up at all in the "active" list.

Anyway glad to see there's evidence of progress, maybe it will just take more time to completely straighten things out.
15) Message boards : Number crunching : Total Credit (Message 53848)
Posted 29 Mar 2016 by jrapdx
Post:
Reading all the comments it certainly looks like the "credit collapse" has gotten a lot of attention. Good thing it's not like bank credit or something, though of course that has happened.

I'm noticing that my CPDN credits haven't changed at all, not up or down in days. What's odd is that trickle uploads are going on, at least the task reports reflect the progress on the tasks quite in sync with BOINC manager on my systems. IOW the work has continued and apparently has been recorded by a server somewhere, at least I think that's the case.

It will be very good to have things working like they're supposed to, the credits are useful because re: serving as feedback that the crunched data isn't lost and probably getting to it's intended destination. I mean I can assume that's occurring since uploads continue, but it is slightly disconcerting that right now it's a matter of guesswork.

Interesting how nervous it's making people feel, like it's a risky situation, though really we individual crunchers out here don't actually have an awful lot at stake.
16) Message boards : Number crunching : Total Credit (Message 53753)
Posted 23 Mar 2016 by jrapdx
Post:
"and has contributed 1,713,935,902,786 Cobblestones of computation
(1,480.84 sextillion floating-point operations) to climateprediction.net"
That is quite funny, the current situation isn't nearly as dramatic but still annoying to people.

FWIW my total hasn't changed even a bit, though it ought to since tasks have been grinding away and lots of trickles have uploaded--successfully by all appearances.

So restoring the "stolen" 100k credits plus the bounty "owed" re: tons of work completed will look pretty impressive when done and said. OK not nearly as far out as sextillions, but a lot more than the total that's reported now.

Any case, not going to hold my breath waiting for it to happen.
17) Message boards : Number crunching : Total Credit (Message 53683)
Posted 18 Mar 2016 by jrapdx
Post:
Between my three linux machines running boinc on wine at the moment, I haven't had any crashes yet. Am inclined to think you are right and it probably is hardware.
So many candidates, disks, video card, motherboard itself, hard to know where to start. One of these days I'll have to swap out components, maybe get some clues.
2nd Edit:works as long as you can type accurately, I typed lower case "H" for?Home!
Thanks for the tip! Yes it does work but not quite so simply. I had to make some modifications, and used this invocation:

/bin/sh -c 'sudo pkill boinc; exec wine "/home/jra/.wine/drive_c/Program Files/BOINC/boincmgr.exe" '

I ran into the problem that if tasks are running when a crash occurs, then on reboot the session manager automatically restarts the tasks. The resumed tasks prevent boincmgr.exe from making connection to the local client, causing boincmgr to error out. Terminating boinc tasks with pkill before calling boincmgr allows it to run normally.

BTW I'm running Ubuntu 15.10 with XFCE. The startup apps are set in the "Session and Startup" dialog. Took a few tries but finally got it right. Thanks again for your help!
18) Message boards : Number crunching : Total Credit (Message 53681)
Posted 17 Mar 2016 by jrapdx
Post:
Like all others in this thread my credits are messed up, 100k short, etc. I'm also not worried about it, seems the servers are down from time to time, anyway this episode isn't nearly as "traumatic" as the outage last fall.

One peculiarity this time is the way credits are "mixed up" among computers. Probably just due to using old data, in which case it will be rectified once things are running normally. Is there a forecast when that will be?

Another thing I've noticed recently is my Linux box episodically rebooting, annoying because Wine/boinc/CPDN doesn't restart automatically, at least I've not been able to get it to do so. I've wondered if some of the unstable CPDN tasks have contributed to these crashes. I'm inclined to think it's more likely a hardware-related issue though not yet been able to pin it down.

The tasks keep chugging along, trickles are uploading so the work is getting done. Looking forward to the moment the credits are finally updated, I'm sure the numbers will be impressive.
19) Message boards : Number crunching : Error while computing (Message 53518)
Posted 25 Feb 2016 by jrapdx
Post:
It's not really clear what the Windows version means for Wine. I've looked at the documentation but difficult to sort out how it affects execution of apps like boinc*. I could change it from Win10 but my hunch is it won't matter. At this point BOINC/CPDN are running OK. Today a task completed and after a new one started 4 tasks are going so I'm inclined to leave things alone for now.

The main problem I've had with Wine is reliably starting BOINC. I was trying to set it up so if the computer reboots (prone to random power outages in this location), BOINC would be automatically restarted. However, despite various attempts with shell scripts, etc., it hasn't worked. The only way it does work requires manually changing to BOINC program directory to start BOINC in a terminal. I need to learn more about the intricacies of Wine, I've hardly used it up to now.

BTW with the Ubuntu PPA Wine version is now 1.9.4, I updated it yesterday.
20) Message boards : Number crunching : Error while computing (Message 53509)
Posted 24 Feb 2016 by jrapdx
Post:
Thank you!

I've tried the trick of opening another tab to load the URL, but it doesn't always work. I mean the indicator will be spinning on that tab instead of this one. Info about the firewall is interesting, how well I can relate, it's sometimes hard getting things to work like we believe they ought to. Anyway, it sheds some light on noticing how slow CPDN has been been lately, obviously the problem I was recently having...


Next 20

©2020 climateprediction.net