Once again, surprise: BOINC doesn't execute any more tasks. The list of running or planned tasks is empty. No more tasks are being downloaded.

It took me a while to discover the reason: This time, my harddisk partition where the projects are stored was filled beyond my given limit of disk space to keep free.

I wish BOINC would list more obvious warnings in the "Messages" window if it detects conditions which prevent it from working at all.

How more obvious? When you have no more room for BOINC to work in, it will tell you that. Then you get messages such as e.g. Message from server: SETI@home Enhanced needs 32.00MB more disk space. You currently have 0.00 MB available and it needs 32.00 MB.

If you want 'more obvious' messages, then please tell how. Give examples of what you think is a better warning. Perhaps that the developers can then add it.

Instead you now complain about it and expect someone else to solve it and come up with a better way of communicating the problem to the user.

---

It would help if BOINC had some way of hiding all the superfluous messages (such as 'restarting task 282-f9f7633k_0-0") in Event Log and displaying only error messages. A button that toggles 'tween All Messages and Error Messages is the general idea. Yes, we can open stdoutdae in a text editor and search for words like error, failed, missing and so on but your average BOINCer doesn't have a clue about stdoutdae, where to find it, what a text editor is, how to open a file in it, let alone how to search for things. They need it simple and easy. What they really need is for someone to remote login to their host and fix whatever's wrong and the sooner the devs implement that the better but until then an errors only window would be a huge improvement.

Another option is to display all error messages in red and warnings in yellow. It probably seems to you that nobody could possibly miss a red line in a sea of black but that's because you have normal to above average reading skills. People with lesser skill will not spot the red line if text is scrolling because the effort required to scroll the text with their mouse/keyboard hand distracts their brain away from the signals their eye is sending to the brain. That red line can go right by them and they won't see it. Hard for someone with your reading skills to imagine that some people are that challenged but they are.

---

"Windows" -- an American English word, meaning "A real operating system is too hard for me."

If you want 'more obvious' messages, then please tell how. Give examples of what you think is a better warning. Perhaps that the developers can then add it.

Instead you now complain about it and expect someone else to solve it and come up with a better way of communicating the problem to the user.

OK, here goes:

"Not requesting tasks: project is not highest priority".
Aside from the missing words which I believe should be "... according to BOINC manager and nothing to do with Resource Share settings", it would be helpful and useful to users to have a method for seeing the current list and some data that helps the user understand why a particular project has a particular priority (according to BOINC manager). How about listing the project priority on the Projects tab?

"Not requesting tasks: scheduler RPC backoff".
I have no suggestion for this because I have no clue what it means. Is it something I can do something about? Or is it a failure at the server end? I'm 100% confident the developers can find a more useful and less jargonistic message.

"This computer has reached a limit on tasks in progress"
I understand what this means but it doesn't tell me why it was displayed. Is there a setting I can adjust? Is there anything I can do about it at all? Where can I go to see what this limit is? Having said that, in every occurrence the application then downloaded a new task and immediately started it so the warning was pointless at best.

"Task <task name> exited with zero status but no 'finished' file"
I'm then warned I may need to reset the project. However, all these tasks restarted automatically and completed and earned credit. I still do not know why this warning was given, nor why I should reset the project, especially as work restarted and completed successfully.

Please let me know if I need to expand on any point here. More than happy to help make the BOINC experience user-friendly to the max.

Thanks!!

doggybob wrote:

...but your average BOINCer doesn't have a clue about stdoutdae, where to find it, what a text editor is, how to open a file in it, let alone how to search for things.

Which is the intent of the Messages window. Before this time it was a Messages tab in BOINC Manager, yet 'the average BOINC user' would perceive all messages in there as error messages, get scared and uninstall BOINC. That doesn't help science at all.

And so there's a somewhat hidden Messages window, for the more advanced types of people, who know a thing or two about their computer and about BOINC. You can adjust the amount of information you see through adding or subtracting flags in the client configuration file cc_config.xml.

As for the actual task related errors, they're easily found at the project web sites. All errors are stored as Errors in the Tasks section of Your Account, the stderr.txt part of the task is stored with every task's Result ID outcome. If you need help then, you can point to such an error in the project forums.

Steve Hawker* wrote:

"Not requesting tasks: project is not highest priority".
Aside from the missing words which I believe should be "... according to BOINC manager and nothing to do with Resource Share settings", it would be helpful and useful to users to have a method for seeing the current list and some data that helps the user understand why a particular project has a particular priority (according to BOINC manager). How about listing the project priority on the Projects tab?

That's correct, the priority at which the next project is chosen for work fetch has nothing to do with resource share.

To see the current list of projects with priorities, enable the <work_fetch_debug> flag in the client configuration file cc_config.xml
Then you get an output similar to this:

16/12/2012 12:21:57 |  | [work_fetch] work fetch start
16/12/2012 12:21:57 |  | [work_fetch] buffer_low: no; sim_excluded_instances 0
16/12/2012 12:21:57 |  | [work_fetch] buffer_low: no; sim_excluded_instances 0
16/12/2012 12:21:57 |  | [work_fetch] ------- start work fetch state -------
16/12/2012 12:21:57 |  | [work_fetch] target work buffer: 86400.00 + 43200.00 sec
16/12/2012 12:21:56 |  | [work_fetch] Request work fetch: Core client configuration
16/12/2012 12:21:57 |  | [work_fetch] work fetch start
16/12/2012 12:21:57 |  | [work_fetch] buffer_low: no; sim_excluded_instances 0
16/12/2012 12:21:57 |  | [work_fetch] buffer_low: no; sim_excluded_instances 0
16/12/2012 12:21:57 |  | [work_fetch] ------- start work fetch state -------
16/12/2012 12:21:57 |  | [work_fetch] target work buffer: 86400.00 + 43200.00 sec
16/12/2012 12:21:57 | Einstein@Home | [work_fetch] REC 63266.431 priority -2.014316
16/12/2012 12:21:57 | SETI@home | [work_fetch] REC 871.905 priority -0.036782
16/12/2012 12:21:57 | Enigma@Home | [work_fetch] REC 114.977 priority -1000.001789
16/12/2012 12:21:57 | Albert@Home | [work_fetch] CPU: fetch share 0.000 rsc backoff (dt 0.00, inc 0.00) (blocked by prefs)
16/12/2012 12:21:57 | Einstein@Home | [work_fetch] CPU: fetch share 0.000 rsc backoff (dt 0.00, inc 0.00) (blocked by prefs)
16/12/2012 12:21:57 | SETI@home | [work_fetch] CPU: fetch share 1.000 rsc backoff (dt 0.00, inc 0.00)
16/12/2012 12:21:57 | CERNVM/Vboxwrapper Test Project | [work_fetch] ATI: fetch share 0.000 rsc backoff (dt 0.00, inc 0.00) (no apps)
16/12/2012 12:21:57 | Einstein@Home | [work_fetch] ATI: fetch share 1.000 rsc backoff (dt 0.00, inc 0.00)
16/12/2012 12:21:57 | SETI@home | [work_fetch] ATI: fetch share 0.000 rsc backoff (dt 0.00, inc 0.00) (blocked by prefs) (no apps)
16/12/2012 12:21:57 | Enigma@Home | [work_fetch] ATI: fetch share 0.000 rsc backoff (dt 0.00, inc 0.00) (no apps)
16/12/2012 12:21:57 |  | [work_fetch] ------- end work fetch state -------
16/12/2012 12:21:57 |  | [work_fetch] No project chosen for work fetch

This one for all your CPU projects and GPU projects separately.

The scheduling priority is thus not based on what your resource share is, but on the amount of work that the projects have done. See Client Scheduler for an in-depth and highly technical write-up.

Technically seen it's not BOINC Manager either that does anything. BOINC consists of two main parts, being the client --that does all the command decisions, cache storing, downloading/uploading/other communications etc.-- and the manager --the graphical user interface, in this case called BOINC Manager, as it manages BOINC. The graphical user interface is used to give easier commands to BOINC, as else you'd have to do all that through the command line and BOINCCMD Tool commands.

"Not requesting tasks: scheduler RPC backoff".
I have no suggestion for this because I have no clue what it means. Is it something I can do something about? Or is it a failure at the server end? I'm 100% confident the developers can find a more useful and less jargonistic message.

And RPC backoff means that there's a timeout going on between your BOINC and the project servers. You can't do anything about it, just wait until it's ended. Any intrusions done by you will result in the RPC resetting and your BOINC having to wait longer.

Yet again you expect someone else to make it explained easier. It's computer jargon that probably cannot be explained easier. Check the link to RPC I gave you.

"This computer has reached a limit on tasks in progress"
I understand what this means but it doesn't tell me why it was displayed. Is there a setting I can adjust? Is there anything I can do about it at all? Where can I go to see what this limit is? Having said that, in every occurrence the application then downloaded a new task and immediately started it so the warning was pointless at best.

Limits are set into place by projects to stop your computer from downloading full caches and then returning only erroneous work. There's no use in your computer getting 10+10 days of cache when all it returns is garbage. So with every erroneous task your limit per CPU per day goes down by a task, up until the point where your computer can only download 1 task per CPU (core) or GPU per day. The rest of that day you will see that error.

What can you do about it? Check why your computer is returning only erroneous work. Clean it, give it better drivers, check that hardware is still intact, etc. When your computer starts returning correct work again, the cache limit will double with each correctly returned task.

Some projects have a maximum of amounts of tasks you can download at the same time, such as Milkyway, it only allows 6 tasks to be downloaded at a time. You will get this same message when your computer requests more work, but it still has the 6 tasks in cache, or at least not reported.

"Task <task name> exited with zero status but no 'finished' file"
I'm then warned I may need to reset the project. However, all these tasks restarted automatically and completed and earned credit. I still do not know why this warning was given, nor why I should reset the project, especially as work restarted and completed successfully.

Yep, that can happen. The science application lost contact with the BOINC client or vice versa, and each time that happens, BOINC will print that message.

It can have been that your anti virus or other anti-malware program was in the way and actively scanning the science application. Therefore we advise to exclude the BOINC Data directory and every file and sub-directory in it in your anti-virus and other anti-malware programs and to only scan these directories by hand after you suspended BOINC, or stopped it.

---

Ageless wrote a lot of stuff so let's break this down into manageable pieces...

Yet again you expect someone else to make it explained easier.

Yes I do. And, this is going to shock you, I am not alone in this respect. There are over 2 million users of BOINC - do you seriously expect them all to have computer science degrees and/or an in-depth understanding of BOINC at a technical level? BOINC does a spectacularly poor job of explaining itself to non-technical people who only want to know how the system works so they can understand and enjoy using the system the way they want to use it.

Is it really such a hardship to explain how the system works?

It's computer jargon that probably cannot be explained easier.

So what you're saying is "Ooooh, it's computer magic that mere mortals cannot understand..."

There is no such thing. There are people who don't want to explain things but there's no such thing as an unexplainable concept.

To see the current list of projects with priorities, enable the <work_fetch_debug> flag in the client configuration file cc_config.xml

I do not have a cc_config file. But it's interesting to see that the priorities are available and therefore BOINC manager is able to retrieve them and then display them.

"This computer has reached a limit on tasks in progress"
I understand what this means but it doesn't tell me why it was displayed. Is there a setting I can adjust? Is there anything I can do about it at all? Where can I go to see what this limit is? Having said that, in every occurrence the application then downloaded a new task and immediately started it so the warning was pointless at best.

Limits are set into place by projects to stop your computer from downloading full caches and then returning only erroneous work. There's no use in your computer getting 10+10 days of cache when all it returns is garbage. So with every erroneous task your limit per CPU per day goes down by a task, up until the point where your computer can only download 1 task per CPU (core) or GPU per day. The rest of that day you will see that error.

What can you do about it? Check why your computer is returning only erroneous work. Clean it, give it better drivers, check that hardware is still intact, etc. When your computer starts returning correct work again, the cache limit will double with each correctly returned task.

Some projects have a maximum of amounts of tasks you can download at the same time, such as Milkyway, it only allows 6 tasks to be downloaded at a time. You will get this same message when your computer requests more work, but it still has the 6 tasks in cache, or at least not reported.

Not exactly sure what you mean by erroneous work but at the time I got that message the only "erroneous" work I generated was a bunch of WUs that I aborted because the project that sent it estimated run times that forced those WUs to run immediately at high priority forcing another project to suspend its WUs and that second project was set up without checkpointing so I lost days of crunching. Not wishing to lose any more days of crunching I aborted the offenders.

Not to make a big deal but I currently have 8 Milkyway WUs that were sent at the same time and I've already crunched at least one that came in with those. Just FYI.

Steve Hawker* wrote:

"Task <task name> exited with zero status but no 'finished' file"
I'm then warned I may need to reset the project. However, all these tasks restarted automatically and completed and earned credit. I still do not know why this warning was given, nor why I should reset the project, especially as work restarted and completed successfully.

Yep, that can happen. The science application lost contact with the BOINC client or vice versa, and each time that happens, BOINC will print that message.

It can have been that your anti virus or other anti-malware program was in the way and actively scanning the science application. Therefore we advise to exclude the BOINC Data directory and every file and sub-directory in it in your anti-virus and other anti-malware programs and to only scan these directories by hand after you suspended BOINC, or stopped it.

I don't have anti-virus. I have a Mac. Never needed any.

Perhaps the message appeared because my router dropped the internet connection? In which case, why don't I see this message every time that happens? I do get the standard message about not being able to connect to the internet but I've only ever seen this zero status message once (well, eight times at once because my CPU has 8 cores)

Is it really such a hardship to explain how the system works?

Yes, as it comes down to either the 3 developers having to do it, and they think too technical, or to one of its users to do it, and then that's you, me, or any of the other 2 million users...

Now you are expecting the 3 developers to go write and dumb it down for you. So why not do it yourself? Why should others do it?

There's a BOINC Wiki here, in which you can go explain all you want. The only thing you need for it is an account, and to stop spammers from filling the Wiki up with crap, the account creation's done by one of the developers only, or in this case David Anderson. So go on email him, request an account for the BOINC User Wiki and explain that you're going to be adding explanations that the mere mortals can read, follow and understand.

I have to say, it'll be more than interesting to see other people than the established ones (3 or 4) writing and adjusting things in the user Wiki.

But why do I have the feeling you won't do that and are still expecting someone else to do it? Not having any experience with how BOINC does its things is not a good enough reason, you can learn all about that. :-)

Not exactly sure what you mean by erroneous work

All returned results that did not validate correctly are erroneous.

---

Now, before everyone thinks that I feel nothing should be changed about BOINC, it isn't that. I know certain things about BOINC should be made easier, better documented, rewritten and I send in a lot of things I find on these and other forums in to the developers, lots of which have made it onto their to-do list.

The real problem here though is that there are only 3 developers busy with BOINC, their time is limited. These people will want to spend time with their family and friends, just as you do, not live and breathe BOINC 24/7. So when they are to change something about the client, is it more important to try to squash some bugs, or that the information messages are dumbed down?

At one point in time it's very probable that the developers stop their development, them splitting off to other venues, and you'll be looking at the last available Berkeley client, its development then resting purely on the shoulders of you, the vast and unpaid community.

Another problem is how eloquent are you going to be? And how long is the translation going to be in other languages? Yes, an error message can become more eloquent, stretch for 3 or more lines in English, but how is that going to be translated in all the languages that BOINC is available in?

But whereas "Not requesting tasks: scheduler RPC backoff" can be easily translated into different languages, perhaps something like "Not requesting asks: the last time that BOINC tried to do this, either the server didn't answer, or your internet had its own troubles, or there was no work available. BOINC is backing off for so many minutes before it tries again automatically. You can stop mashing the Update button." is a bit more difficult to translate. ;-)

If there was money enough at Berkeley, if they had programmers available to them, if there were enough people willing to do research, learn and document, I'm sure it would all be possible. Programmers and documenters aren't going to help out if they aren't going to be paid for it. Everyone always feels others will have to do it for them, if that makes it easier to them.

Then to say "Yeah, but I don't know anything about the program", just isn't good enough. You'll get the answer, "Why not go and try to learn about it? It's what you expect of someone else to do as well, as how else is the other person going to make the program easier to understand?"

---

doggybob wrote:

...but your average BOINCer doesn't have a clue about stdoutdae, where to find it, what a text editor is, how to open a file in it, let alone how to search for things.

Which is the intent of the Messages window. Before this time it was a Messages tab in BOINC Manager, yet 'the average BOINC user' would perceive all messages in there as error messages, get scared and uninstall BOINC. That doesn't help science at all.

And so there's a somewhat hidden Messages window, for the more advanced types of people, who know a thing or two about their computer and about BOINC. You can adjust the amount of information you see through adding or subtracting flags in the client configuration file cc_config.xml.

We were talking about how the error messages in Event Log could be made more visible weren't we? I suggested a button that would temporarily hide all the non-error messages and show only error messages. While your response is certainly factual I don't think it relates to my suggestion. Or maybe I missed the point. Please explain.

As for the actual task related errors, they're easily found at the project web sites. All errors are stored as Errors in the Tasks section of Your Account, the stderr.txt part of the task is stored with every task's Result ID outcome. If you need help then, you can point to such an error in the project forums.

That's all true but doesn't explain why my suggestion (which was to hide all the non-error messages and show only the error messages to make them easier to find since then they would not be mixed in with all the non-error messages) is a good idea or a bad idea. That was the OP's complaint: the error messages are hard to pick out when they are mixed in with all the other messages. He wasn't asking for a better way to know what the messages mean, he was just asking for something that would make them easier to spot in Event Log. Again, you may not be able to comprehend that some people would find it difficult to spot an error message and that's probably because you have average to above average reading skill but many,many readers are not so lucky. Again, this has nothing to do with interpreting the messages, it all about making the messages themselves easier to see in Event Log. Maybe now you understand better why I mentioned loading the entire log into a text editor and using the editor's search function to look for words like error, missing, not found, etc. as that would help a person find the lines that contain an error message.

---

"Windows" -- an American English word, meaning "A real operating system is too hard for me."

Their budget is also dwindling fast, the grands money being less each year.

Grants, not grands. <===== nit pick

That problem is easy to solve. Berkeley is an American university. All the Americans need to do is whip up some lies about some WMD, invade some helpless country over the lies, borrow all the money they need to fun the invasion and then poof! the Magic Money Genie will make them all incredibly wealthy and happy. There will be a pot on every stove and a plump chicken in every pot. The BOINC devs will have more money than they can possibly spend in 2 lifetimes. They'll hire all the programmers who don't have jobs now and they'll churn out code faster than an Indian pukes up Listerine the morning after his welfare cheque goes into the bank.
</sarcasm>

At one point in time it's very probable that the developers stop their development, them splitting off to other venues, and you'll be looking at the last available Berkeley client, its development then resting purely on the shoulders of you, the vast and unpaid community.

Oh happy day! The first thing we'll do is snip out all the credits related code which will put an end to the barrels of nonsense and wasted time that happens due to credits. Then we'll base the scheduler on the realities of the modern world instead of pretending we all run 300 baud phone line modems we bought in 1984. (What purpose does a cache serve other than to inflate SETI's database to the point their server can't handle it? Oh, it's so we can grab extra tasks in case their server goes down due to database overload. Wow, that's really smard!!!)

Another problem is how eloquent are you going to be? And how long is the translation going to be in other languages? Yes, an error message can become more eloquent, stretch for 3 or more lines in English, but how is that going to be translated in all the languages that BOINC is available in?

Not a big problem. After the Americans realize how rich they can become destroying helpless countries on borrowed money they will invade *everybody* and make them *all* speak English. That will put the people programming the online translators out of work but of course Berkeley will be able to hire them and churn out BOINC code even faster. Don't worry, once the lies about WMD get rolling and the troops get pumped up it will all fall into place.

But whereas "Not requesting tasks: scheduler RPC backoff" can be easily translated into different languages, perhaps something like "Not requesting asks: the last time that BOINC tried to do this, either the server didn't answer, or your internet had its own troubles, or there was no work available. BOINC is backing off for so many minutes before it tries again automatically. You can stop mashing the Update button." is a bit more difficult to translate. ;-)

Nonsense. Difficult for an online translator bot but not difficult for a human fluent in both languages.

The real problem is Berkeley's stoneage thinking. They don't want to add another 100 KB to BOINC package because they still think we all run 300 baud modems connected to our phone line. They can't get past that so turf 'em, sooner the better.[/quote]

If there was money enough at Berkeley, if they had programmers available to them, if there were enough people willing to do research, learn and document, I'm sure it would all be possible. Programmers and documenters aren't going to help out if they aren't going to be paid for it.

Nonsense. The problem is Berkeley does nothing to curry the favor of those who can help and instead throws every possible roadblock and stoneage mentality in their way. Little wonder they get no help.

---

"Windows" -- an American English word, meaning "A real operating system is too hard for me."

doggybob wrote:

I suggested a button..

David Anderson wrote:

No need to discuss extensions to the messages mechanism.

LigH wrote:

more obvious warnings

David Anderson wrote:

We have a mechanism for informing the user of conditions that require their attention: notices.

So, the out-of-disk condition should get a notice.

---

So, the out-of-disk condition should get a notice.

That works for me for the out-of-disk problem but I don't think it's the best solution for the problem in general and that problem is: errors and warnings are difficult to see when they are included in a deluge of messages the way they are now. There was and still is a very simple solution for that problem and that very simple solution is the one I still propose: a toggle that will hide all the superfluous messages in Event Log and show only errors and warnings. Any message with the word error or warning in it would be shown in that short list therefore all one need to do is put one of those keywods in their error or warning and poof! like magic there error/warning become easy to spot. Yah, I know, that's so simple it sucks. There must be a more complicated way and once again Dave has shown he is the man who can find that way.

Instead of implementing the simple, easy, direct solution he creates a new problem by creating Notices. Now we have to fight and argue over each and every important condition/message that should get a notice, for example the out-of-disk notice. Now us lucky people have earned the privilege of fighting to have all the other errors put in notices too. We'll have to fight message-by-message, condition-by-condition, each one a separate fight and argument, the whole thing shouldn't take more than 27 years as it gets put on the back burner so more newly screwed up stuff (like disk sharing) can be implemented first. He doesn't fix problems, he just turns them into new and more numerous problems and does so in ways that make the basic problem even more convoluted and harder to fix. That's not software engineering. That's called "screwing stuff up and turning half-broken easily fixed things into a completely broken hard to fix tangled mess". That's more truth than he can swallow so I see a banishing in my near future. Lol, as if I care. As if there aren't a trillion other IPs to post from.

---

"Windows" -- an American English word, meaning "A real operating system is too hard for me."

Hmmm, I seem to remember that BOINC is something David came up with, not you. It is however an Open Source program, you can go and download the source code, change everything you don't like, add to it and then release it under a name of your own.

---