Message boards : Projects : WCG OPNG sans OPN1
Message board moderation
Author | Message |
---|---|
Send message Joined: 13 Sep 17 Posts: 26 |
Since WCG made the mistake of sending both GPU and CPU WUs down the same "pipe," as KU put it, I don't understand the correct way to specify what I want in the Device Profile. I now would like to focus on ARP, HST and MCM plus OPNG. But it seems that if I set Allow research to run on my CPU? Yes and Project Limits OpenPandemics - COVID-19 = 1 then I only get one OPN1 and one OPNG. I want 50 OPNG WUs so do I have set Project Limits OpenPandemics - COVID-19 = 50??? I'd ask in the WCG forum but IBM shadow-banned me. |
Send message Joined: 8 Nov 10 Posts: 310 |
We had a long discussion on that right at the beginning of OPNG. You can't separate the CPU and GPU work units, so if you want 50, you will get 50. I think that is total, though they probably won't send you enough GPU work anyway, so you will end up doing the CPU stuff. It is a waste of everyone's time and resources, but that is the way they do it. (I do Folding.) |
Send message Joined: 28 Jun 10 Posts: 2691 |
ARP is CPU only. |
Send message Joined: 27 Jun 08 Posts: 641 |
We had a long discussion on that right at the beginning of OPNG. You can't separate the CPU and GPU work units, so if you want 50, you will get 50. One one of my NVidia systems, I set the venue "no cpu" and "allow nvidia" but have not received any tasks from WCG. On a linux ATI system that is open for CPU and GPU I get a boatload of CPU tasks and a dinghy load of GPU. I was guessing there are no NVIdia tasks available but a few ATI. Maybe the problem is linux ATI but no windows NVidia ??? Or is the problem getting the NVidia because I specified no CPU? I poked around WCG but cannot find a list of applications. Their site is so different from other projects it is difficult to even find my own account. Is there a list of apps or even a server status page? |
Send message Joined: 8 Nov 10 Posts: 310 |
One one of my NVidia systems, I set the venue "no cpu" and "allow nvidia" but have not received any tasks from WCG. (1) You need to specify CPU in order to get GPU. And they have the same pile of GPU work that they send either to Nvidia or ATI (they are both OpenCl), so when they have work for one, they have work for the other. And when they are out (more likely), they are out for both. (2) I am sure that they don't have a server status page; that is a known shortcoming, and I have never seen an app page either. I believe they were doing their own thing before BOINC, and never entirely adapted. |
Send message Joined: 24 Dec 19 Posts: 229 |
did they change this? because it was not like that before. I have all of my systems selected for only OP-COVID work, with CPU work disallowed, and GPU allowed on Nvidia. I had no trouble getting work this way. |
Send message Joined: 5 Oct 06 Posts: 5128 |
Somebody may be mis-remembering the problem we had (and may still have) getting work for intel GPUs without also specifying something else. I'm getting work for both intel GPUs and NVidia GPUs, sometimes in the same fetch, but without any work for the CPU. |
Send message Joined: 30 Mar 20 Posts: 416 |
I get both Nvidia and iGPU tasks without having set the choice for CPU to "Yes". Not that it's easy to get GPU work, but one certainly do not have to set CPU to "Yes". |
Send message Joined: 8 Nov 10 Posts: 310 |
Yes, I misstated that. Aurum said that he wanted the other CPU projects, so he would have to set CPU to "yes". But then he would get the CPU version of OPN also. If you don't want any CPU projects, then you don't have to set CPU to yes. |
Send message Joined: 17 Nov 16 Posts: 890 |
I poked around WCG but cannot find a list of applications. Their site is so different from other projects it is difficult to even find my own account. Is there a list of apps or even a server status page? On the old IBM hosted website, both apps and stats were available just like all other BOINC projects. But they never had those listed in the menus. You had to know the URL of the pages and input that yourself like you have to do with other similar websites that don't show any menus options. I have to to do the same for GPUGrid for example. You just pull up the normal server status page for example and replace stats with apps in the URL and it will show you the applications that are available for the project. It still used the basic BOINC server code that produces the normal set of pages. But apparently you can choose to not make them publicly available in the stock menus. But this new Krembil website is so different it is hard to find anything. |
Send message Joined: 25 May 09 Posts: 1301 |
You are about a month behind the times. They have a new website, neither designed nor hosted by IBM (but I dare say there is some IBM influence lurking somewhere). There's new Ts&Cs to read, new privacy statements to get to grips with and a new layout. All the major public facing pages don't refer to IBM (in the current tense), but I dare say some of the applications do. I would assume that the historic server stats are a low priority to be transferred, but given the project(s) are now being hosted on "new" servers that is of no surprise. |
Send message Joined: 13 Sep 17 Posts: 26 |
We had a long discussion on that right at the beginning of OPNG. You can't separate the CPU and GPU work units, so if you want 50, you will get 50.Actually Keith said he made that choice because it made their job easier until it didn't. It was a deliberate choice not a "have to." |
Send message Joined: 13 Sep 17 Posts: 26 |
Yes, I misstated that. Aurum said that he wanted the other CPU projects, so he would have to set CPU to "yes".Exactly. In order to run MCM and ARP I have to enable "use my CPU." That sends me tons of OPN1 WUs and gums up the works and makes me abort many OPN1 WUs. Once they even asked us to lighten up on OPN1s because they had more OPNG and couldn't balance it. Trying to alleviate a problem of their own making. If I only want OPNG from WCG I could disable using my CPU and run CPU WUs from a different project. I was imploring Keith Uplinger not to do it that way but he did. Hopefully KRI will never make that mistake again. |
Send message Joined: 13 Sep 17 Posts: 26 |
Last 2 days I'm having a huge backlog of WUs trying to upload. The forums seem to say the problem's gone but as of now it's still a big problem. 28618 10/29/2021 8:57:33 AM Project communication failed: attempting access to reference site 28619 World Community Grid 10/29/2021 8:57:33 AM Temporarily failed upload of OPN1_0084321_00498_0_r1080501240_0: transient HTTP error 28620 World Community Grid 10/29/2021 8:57:33 AM Backing off 00:05:28 on upload of OPN1_0084321_00498_0_r1080501240_0Might this script be exacerbating the UL problem by putting me on a badboy list??? #!/bin/sh cd /var/lib/boinc watch -n 164 "sudo boinccmd --project 'http://www.worldcommunitygrid.org/' update | sudo boinccmd --network_available"Glad to see they added a 200-year badge but they also need 500 and 1000 years. |
Send message Joined: 5 Oct 06 Posts: 5128 |
Set 'http_debug' in the Event Log options, and find out what that 'transient HTTP error' is. |
Send message Joined: 13 Sep 17 Posts: 26 |
What and where am I looking for the http_debug output??? My BoincTasks Messages page is flying with entries like these: 10/29/2021 10:31:23 AM [http] [ID#286] Info: Found bundle for host upload.worldcommunitygrid.org: 0x55650ac53300 [can multiplex] 1196 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Info: Multiplexed connection found! 1197 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Info: Re-using existing connection! (#123) with host upload.worldcommunitygrid.org 1198 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Info: Using Stream ID: 93 (easy handle 0x55650a6446f0) 1199 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Sent header to server: POST /boinc/wcg_cgi/file_upload_handler HTTP/2 1200 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Sent header to server: Host: upload.worldcommunitygrid.org 1201 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Sent header to server: user-agent: BOINC client (x86_64-pc-linux-gnu 7.16.6) 1202 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Sent header to server: accept: */* 1203 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Sent header to server: accept-encoding: deflate, gzip, br 1204 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Sent header to server: accept-language: en_US 1205 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Sent header to server: content-length: 17059553 1206 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Sent header to server: content-type: application/x-www-form-urlencoded 1207 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Sent header to server: 1208 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#286] Sent header to server: Ä£Æà»-¿È×…@ 1209 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Info: Found bundle for host upload.worldcommunitygrid.org: 0x55650ac53300 [can multiplex] 1210 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Info: Multiplexed connection found! 1211 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Info: Re-using existing connection! (#123) with host upload.worldcommunitygrid.org 1212 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Info: Using Stream ID: 95 (easy handle 0x556507c6bb30) 1213 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Sent header to server: POST /boinc/wcg_cgi/file_upload_handler HTTP/2 1214 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Sent header to server: Host: upload.worldcommunitygrid.org 1215 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Sent header to server: user-agent: BOINC client (x86_64-pc-linux-gnu 7.16.6) 1216 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Sent header to server: accept: */* 1217 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Sent header to server: accept-encoding: deflate, gzip, br 1218 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Sent header to server: accept-language: en_US 1219 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Sent header to server: content-length: 17206908 1220 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Sent header to server: content-type: application/x-www-form-urlencoded 1221 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#287] Sent header to server: 1222 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Info: Found bundle for host upload.worldcommunitygrid.org: 0x55650ac53300 [can multiplex] 1223 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Info: Multiplexed connection found! 1224 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Info: Re-using existing connection! (#123) with host upload.worldcommunitygrid.org 1225 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Info: Using Stream ID: 97 (easy handle 0x55650976cb90) 1226 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Sent header to server: POST /boinc/wcg_cgi/file_upload_handler HTTP/2 1227 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Sent header to server: Host: upload.worldcommunitygrid.org 1228 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Sent header to server: user-agent: BOINC client (x86_64-pc-linux-gnu 7.16.6) 1229 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Sent header to server: accept: */* 1230 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Sent header to server: accept-encoding: deflate, gzip, br 1231 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Sent header to server: accept-language: en_US 1232 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Sent header to server: content-length: 285 1233 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Sent header to server: content-type: application/x-www-form-urlencoded 1234 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Sent header to server: 1235 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#288] Sent header to server: ç<8)j/å‘UÿìoXÕ-ÉÞLfq×Úæã½)¨úá[¹ƒ¨AH¼qçbå-²ëß|ÏÛ«ZÊ1ñL"ó÷‚ùsy¾×KÀÑ<“AdÑæ¾#" 1236 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Info: Found bundle for host upload.worldcommunitygrid.org: 0x55650ac53300 [can multiplex] 1237 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Info: Multiplexed connection found! 1238 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Info: Re-using existing connection! (#123) with host upload.worldcommunitygrid.org 1239 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Info: Using Stream ID: 99 (easy handle 0x55650a6990f0) 1240 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Sent header to server: POST /boinc/wcg_cgi/file_upload_handler HTTP/2 1241 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Sent header to server: Host: upload.worldcommunitygrid.org 1242 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Sent header to server: user-agent: BOINC client (x86_64-pc-linux-gnu 7.16.6) 1243 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Sent header to server: accept: */* 1244 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Sent header to server: accept-encoding: deflate, gzip, br 1245 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Sent header to server: accept-language: en_US 1246 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Sent header to server: content-length: 20209328 1247 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Sent header to server: content-type: application/x-www-form-urlencoded 1248 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Sent header to server: 1249 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Sent header to server: 0e€òÀà‚C>ûî/g2–5\Û¸ËСI¤G0E10 UBM10U 1250 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#289] Sent header to server: QuoVadis Limited10UQuoVadis Root CA 3‚± 1251 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#290] Info: Found bundle for host upload.worldcommunitygrid.org: 0x55650ac53300 [can multiplex] 1252 World Community Grid 10/29/2021 10:31:23 AM [http] [ID#290] Info: Multiplexed connection found! |
Send message Joined: 5 Oct 06 Posts: 5128 |
Knock off the debug flag now - they'll all be the same. I'll look through this one. [edit] - ouch - how many files do you have uploading at once? Each one only just seems to get started, but then attention switches to another one. For this sort of job, it's easier to wait until every file is in backoff, then set the http_debug flag, retry just one file, and knock the flag off again. Then you get something like 29/10/2021 18:58:24 | World Community Grid | Computation for task OPNG_0098417_00141_0 finishedYours seems to fail at the one I've marked, but without seeing one the whole way through - from 'started upload' to 'finished upload' - it's hard to be sure. I'll go and look at a Linux machine, and see how it compares. |
Send message Joined: 13 Sep 17 Posts: 26 |
Hmm, scratches head... BoincTasks doesn't have a Suspend Uploads. I'll have to see if I can open BOINCmgr but it usually says "Disconnected." And it is. There's 62 files trying to UL on Rig-44 alone and it's the worst actor. I have hadam4h and ARP WUs running and it could be 6 hours before ARP checkpoints. If I suspend from the BoincTasks task tab it switches ULs to upload Pending. No, that was just a project backoff to 5 hours. ARP checkpointed so I'll suspend everything but one and reboot. |
Send message Joined: 5 Oct 06 Posts: 5128 |
Tried one on a Linux (Mint v20.2) machine. It starts... 29/10/2021 19:24:35 | World Community Grid | Started upload of OPNG_0098436_00093_1_r329524617_0- I won't bore you with the rest. Seems like the multiplexing might be your problem, but I don't know Linux well enough to go much further. |
Send message Joined: 13 Sep 17 Posts: 26 |
This from my cc_config: <fetch_minimal_work>0</fetch_minimal_work> <fetch_on_update>1</fetch_on_update> <force_auth>basic</force_auth> <http_transfer_timeout>3000</http_transfer_timeout> <http_transfer_timeout_bps>10</http_transfer_timeout_bps> <http_1_0>0</http_1_0>Should fetch on update be set??? I wish there were better explanations for all these options. |
Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License,
Version 1.2 or any later version published by the Free Software Foundation.