Message boards :
Number crunching :
WU abuse
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 . . . 10 · Next
Author | Message |
---|---|
Send message Joined: 12 Mar 09 Posts: 61 Credit: 9,214,340 RAC: 0 |
Not sure who you are directing this towards, but if it is towards me you got it I am using an official BOINC client, although an old one (5.8.16). That client does not have the capability of reporting immediately. yes it has, you can set it in the cc_config.xml As I explained in another post, my tasks are reporting as they complete because my cache setting is equal to the length of the deadline. regardless of how you managed this behavior, compared to it, update scripts are peanuts BTW, dual and quad core systems probably would naturally report more tasks at once... MW reports as many WUs as it got with the last download if it isn't tricked to behave another way |
Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0 |
Prior to having a communication weirdness with Einstein@Home, I did not even have a cc_config.xml. As it stands right now, all I have is a cc_config.xmi, basically a rename to where I have what I did so that if I need to enable the feature that I used over there, I don't have to go back and figure it out... Here's the contents of that file: <cc_config> <log_flags> <http_debug>1</http_debug> <http_xfer_debug>1</http_xfer_debug> </log_flags> </cc_config>
Convenient that you snipped out the FACT that I had tried to rectify the behavior because I knew the impact. When I tried to rectify it, it made it worse. I was making constant requests for work every 1 minute, working on up to 7-10 minutes. I cannot make my system report more than one at a time due to the low per-core limit. You go ahead and throw that back at me if you wish, but at least once I realized that I was communicating with the server more often at the lower setting, I bumped back up. BTW, dual and quad core systems probably would naturally report more tasks at once... No, it downloads the limit and then realizes that it might not be able to report back before the task is past deadline and automatically reports. Most anyone who knows anything about the scheduler and work fetch policy will know that... Beyond that, we're done... |
Send message Joined: 12 Mar 09 Posts: 61 Credit: 9,214,340 RAC: 0 |
Sorry, i mixed things up with Pwrguru, he's reporting WUs every few seconds immediately when they're done which i a pain for the servers and completely senseless. You're reporting a WU every 30 Minutes which is perfect Anyway i don't understand your problem with deadlines, they're at 3 days you can do the max 6 WUs in 3 hours so it should be possible to do the 6 WUs and report them back. Not that i think that is needed but it should be possible. kindly regards |
Send message Joined: 22 Nov 08 Posts: 136 Credit: 319,414,799 RAC: 0 |
My computers are not hidden. You can view them anytime you like. The only one running the update script is the q6600. It is the only one that runs out of work. Cache is now set to 3.0 days and task scheduler(auto update script) is now disabled. I'll give it the weekend to see if there is any improvement. 4870 GPU 4870 GPU |
Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0 |
Sorry, i mixed things up with Pwrguru, he's reporting WUs every few seconds immediately when they're done which i a pain for the servers and completely senseless. Apology accepted... Out of curiosity, I have tried changing cache settings down again. I set to 0.04 days and BOINC didn't report one of the tasks after it finished. I think whether or not I can report more than one at a time is dependent upon which type of tasks I get. The 16 minute tasks that I have now, I think once I get down to the point that I have 2 tasks in queue I'll start requesting work again. A work request also reports tasks, unless there is something in one of the config files that can change that as well... So, once I get 2 tasks in queue, I'll be reporting 1 and requesting 1...which goes back to the heavier SQL load. If that's what happens, I still think it is lighter on the server if I still report each task as it completes and let the work request happen once I've run dry, as either way has the same SQL load... |
Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0 |
My computers are not hidden. You can view them anytime you like. The only one running the update script is the q6600. It is the only one that runs out of work. Thanks... |
Send message Joined: 30 Aug 08 Posts: 24 Credit: 250,053,699 RAC: 0 |
Sorry, i mixed things up with Pwrguru, he's reporting WUs every few seconds immediately when they're done which i a pain for the servers and completely senseless.If you read it right I am not reporting a work unit every second as you say.....It is completed work unit uploaded (no reporting), not asking for more work.....I realize that you may be new and all but that is how boinc works..... P.s. I think that you are still confused who you are talking about......... |
Send message Joined: 24 Dec 07 Posts: 1947 Credit: 240,884,648 RAC: 0 |
Sorry, i mixed things up with Pwrguru, he's reporting WUs every few seconds immediately when they're done which i a pain for the servers and completely senseless.If you read it right I am not reporting a work unit every second as you say.....It is completed work unit uploaded (no reporting), not asking for more work.....I realize that you may be new and all but that is how boinc works..... No, BOINC doesn't work like that. The times I posted of your contacts with the server were from wu's being reported. Much more often than anything my update script does. |
Send message Joined: 6 Apr 08 Posts: 2018 Credit: 100,142,856 RAC: 0 |
Sorry, i mixed things up with Pwrguru, he's reporting WUs every few seconds immediately when they're done which i a pain for the servers and completely senseless.If you read it right I am not reporting a work unit every second as you say.....It is completed work unit uploaded (no reporting), not asking for more work.....I realize that you may be new and all but that is how boinc works..... Still trying to defend your update script? Tell that to the forum administrator, Dave, who just said that the server can't keep up with the ATI cards. And before you start with your suggestive lies - I would NEVER use your update script or ANYTHING of yours. Not as if I could since you did't share this script. Sniff. |
Send message Joined: 30 Aug 08 Posts: 24 Credit: 250,053,699 RAC: 0 |
Keep wagging that tail to try and take the focus off yourself....Using my list that you posted, I completed 40 work units in 17 minutes.........Now for some reason that doesn't seem excessive to me since I am running a 4870 1 gig Gpu card and 4 WCG tasks at the same time.........And when I run out of work I wait to get more work....I see that you somehow FORGOT to post that part of my list....... |
Send message Joined: 31 Mar 08 Posts: 61 Credit: 18,325,284 RAC: 0 |
My computers are not hidden. You can view them anytime you like. The only one running the update script is the q6600. It is the only one that runs out of work. Mine is set at 10 days and it makes no difference! On other projects I'm overfilled with tasks! |
Send message Joined: 24 Dec 07 Posts: 1947 Credit: 240,884,648 RAC: 0 |
Congratulations, that is a nice card. Mine is just a 4850 with 512MB and I have 5 other projects working on my quad core, but that is not the point is it. The point is the number of times your computer contacts MW. You accused the folks who use an update script of causing problems with the server, but your system contacts the server way more often than anything my update scrpit is configured to do. That is all I am pointing out. Not distracting the discussion by pointing anything else out. |
Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0 |
I now seem to be alternating between reporting 1 and reporting 2 and only keeping 3 non-processed tasks now... Marginally better every now and then when it comes to db updates, but I'm still asking for work more frequently, so I have more scheduler contacts this way... On the whole it may be a wash... |
Send message Joined: 14 Mar 09 Posts: 16 Credit: 35,833,011 RAC: 0 |
Still trying to defend your update script? Tell that to the forum administrator, Dave, who just said that the server can't keep up with the ATI cards. Dave: But I'm almost certain that the server can't keep up with all the requests from the users with graphics cards. Oh how often I was "almost certain" just to be proven wrong afterwards ;-) I'm sure they will take care of, because I also think that they have interest in faster WU-computation and don't like if the computation stuck just because their server is not able to deliver fast enough. Regards, Lothar |
Send message Joined: 12 Nov 07 Posts: 2425 Credit: 524,164 RAC: 0 |
Still trying to defend your update script? Tell that to the forum administrator, Dave, who just said that the server can't keep up with the ATI cards. I am thinking that Dave ment in general even without the update scripts. Doesn't expecting the unexpected make the unexpected the expected? If it makes sense, DON'T do it. |
Send message Joined: 12 Aug 08 Posts: 253 Credit: 275,593,872 RAC: 0 |
Still trying to defend your update script? Tell that to the forum administrator, Dave, who just said that the server can't keep up with the ATI cards. More cpu time than task's, and this project is still in alpha! |
Send message Joined: 15 Jan 09 Posts: 169 Credit: 6,734,481 RAC: 0 |
I don't see what is wrong with a carefully thought out script. A script that is written to hammer the server more than it needs to is something else. On both my amd dual-core and my Core 2 Duo Ihave had MW delays of up to 24 hours. I dont have a GPU (yet! Look out boys :) ) I run a script that controls a number of projects, and uploads completed workunits every 40 minutes (all have something to upload on most connection attempts). 40 minutes is often enough that if the server does not give work, I won't run out before the next connection. If I knew how to detect whether a specific project is waiting to upload or has run out of work, or better still how to tell how many tasks are in my queue waiting to start, I would write the sriptto check every so often whether a project is running short of work, and I would only connect at that time. I figure connecting every 40 minutes (time on MW and some other projects for doing more than 2 tasks) is not excessive, given that it helpsto prevent myfarm from running out of work on those projects that are most important to me. And yes, I AM a credit "whore". I just don't have the hardware yet to reach the top 1000 here or in BOINC combined :) |
Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0 |
After leaving it alone for 12 hours, it is now mostly doing what I expected, reporting 1 and downloading 1 because of more frequent contacts to the scheduler. My system has reported 2 tasks at the same time only once in the past 12 hours. So to cut back on scheduler connect attempts, I'm switching back to a 3 day cache... |
Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0 |
I don't see what is wrong with a carefully thought out script. A script that is written to hammer the server more than it needs to is something else. I view the problem with scripts as more of the potential for bad than anything else. Sure, some folks may be perfectly responsible, but there will be those out there who are not. Also those who are attempting to be perfectly responsible have the capability of making a mistake, or chosing a setting that one would think is conservative enough but since they don't know what the actual condition of the server is, it might be enough to cause harm. My view might be "different" on this due to my shifting role in my job from a programmer to a system admin. As for the delays here, I'm also curious as to the results of KWSN imcrazynow's system and how it reacted to my suggestion of increasing the cache to 3 days. That should've slowed down the frequency of work requests and, in theory, decreased the chances of getting a long backoff, which hopefully keeps a more steady stream of work... |
Send message Joined: 22 Nov 08 Posts: 136 Credit: 319,414,799 RAC: 0 |
I don't see what is wrong with a carefully thought out script. A script that is written to hammer the server more than it needs to is something else. I've already run out of work 4 times that i'm aware of including right now. From what I see the slower systems have no problems running out of work. My other two never ran out of work with a .5 day cache. I don't expect them to with a 3 day cache either. My quad on the other hand is running out very frequently regardless of the cache setting. It may be contacting the server less but that's not helping out on my end a bit. The server knows how much work you have on hand to be done. Why would it send no work when your cache is gone and there's no work in progress If your cache is gone and you're requesting work you should get work not a back off unless the server is out of work to send. I look at the server status page and see hundreds of tasks to send but I get the big 0 and another back off. <---not very happy camper 4870 GPU 4870 GPU |
©2024 Astroinformatics Group