After watching this for about a week, I have to ask.

Is anybody else having problems getting work?

Is see this in BOINC messages:

SETI@home 11/7/2018 7:10:10 PM update requested by user
SETI@home 11/7/2018 7:10:10 PM Sending scheduler request: Requested by user.
SETI@home 11/7/2018 7:10:10 PM Requesting new tasks for AMD/ATI GPU
SETI@home 11/7/2018 7:10:13 PM Scheduler request completed: got 0 new tasks
SETI@home 11/7/2018 7:10:13 PM No tasks sent
SETI@home 11/7/2018 7:10:13 PM No tasks are available for AstroPulse v7
SETI@home 11/7/2018 7:10:13 PM No tasks are available for SETI@home v8

According to the server stats there is plenty of work, and noone else seems to be having problems getting work.

I havent returned any bad results lately.

I really cant figure out why I'm not getting work.

Its only for the GPU that no work is fetched, the CPU got 100 task some 5 days ago.

http://setiathome.berkeley.edu/results.php?hostid=862199

---

You are not getting any gpu work because you are not asking for it. Need to figure out why. Does BOINC see your graphics card with both the AMD driver and the OpenCL driver loaded? Shown in the beginning of the Event Log when BOINC first started.

Also if you set sched_op_debug in the Event Log Options via the Manager interface, the next scheduled server request for work will print out exactly how many seconds of cpu work and how many seconds of gpu work you are requesting.

Lots of posts lately about people saying they are not getting work. 100% of the reason is they are overcommitted to other projects debt so Seti is last on the list to get any love. So how may other projects are you asking for work for? Reduce the project resources of the other projects and increase Seti's resources.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Thanks for your reply :)

I'm definetely asking for work.

When I'm not, my Boinc shows something like this:

11/7/2018 20:52:30 | Not requesting tasks: don't need (CPU: job cache full; AMD/ATI GPU: not highest priority project)

Even if I manually set all my other projects to no new work, and update, I dont get work.

I'm not a newbie at Boinc, have been with the project since day one.

Boinc sees the GPU:

07-11-2018 21:32:30 | | OpenCL: AMD/ATI GPU 0: Radeon (TM) RX 480 Graphics (driver version 2686.5, device version OpenCL 2.0 AMD-APP (2686.5), 8192MB, 8192MB available, 6244 GFLOPS peak)

And it also gets used for other projects.

I have added the sced_op_debug, and restarted boinc.

Then tried refeshing Seti.

07-11-2018 21:36:44 | SETI@home | Requesting new tasks for AMD/ATI GPU
07-11-2018 21:36:44 | SETI@home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices
07-11-2018 21:36:44 | SETI@home | [sched_op] AMD/ATI GPU work request: 346244.03 seconds; 1.00 devices
07-11-2018 21:36:46 | SETI@home | Scheduler request completed: got 0 new tasks
07-11-2018 21:36:46 | SETI@home | [sched_op] Server version 709
07-11-2018 21:36:46 | SETI@home | No tasks sent
07-11-2018 21:36:46 | SETI@home | No tasks are available for AstroPulse v7
07-11-2018 21:36:46 | SETI@home | No tasks are available for SETI@home v8
07-11-2018 21:36:46 | SETI@home | Project requested delay of 303 seconds

Its quite obvious that I'm asking for work, quite a lot of it (over 96 hours anyway).

Why am I not given any?

I find it curious that the server says no tasks available....

I think I'll try backing up my app_info.xml, and aplication files, and then reset the seti project. Then I'll see if it helps?

---

Thanks for your reply :)

I'm definetely asking for work.

When I'm not, my Boinc shows something like this:

11/7/2018 20:52:30 | Not requesting tasks: don't need (CPU: job cache full; AMD/ATI GPU: not highest priority project)

Even if I manually set all my other projects to no new work, and update, I dont get work.

I'm not a newbie at Boinc, have been with the project since day one.

Boinc sees the GPU:

07-11-2018 21:32:30 | | OpenCL: AMD/ATI GPU 0: Radeon (TM) RX 480 Graphics (driver version 2686.5, device version OpenCL 2.0 AMD-APP (2686.5), 8192MB, 8192MB available, 6244 GFLOPS peak)

And it also gets used for other projects.

I have added the sced_op_debug, and restarted boinc.

Then tried refeshing Seti.

07-11-2018 21:36:44 | SETI@home | Requesting new tasks for AMD/ATI GPU
07-11-2018 21:36:44 | SETI@home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices
07-11-2018 21:36:44 | SETI@home | [sched_op] AMD/ATI GPU work request: 346244.03 seconds; 1.00 devices
07-11-2018 21:36:46 | SETI@home | Scheduler request completed: got 0 new tasks
07-11-2018 21:36:46 | SETI@home | [sched_op] Server version 709
07-11-2018 21:36:46 | SETI@home | No tasks sent
07-11-2018 21:36:46 | SETI@home | No tasks are available for AstroPulse v7
07-11-2018 21:36:46 | SETI@home | No tasks are available for SETI@home v8
07-11-2018 21:36:46 | SETI@home | Project requested delay of 303 seconds

Its quite obvious that I'm asking for work, quite a lot of it (over 96 hours anyway).

Why am I not given any?

I find it curious that the server says no tasks available....

I think I'll try backing up my app_info.xml, and aplication files, and then reset the seti project. Then I'll see if it helps?

The line highlighted in red shows why. Not highest priority project. Even though the sched_op_debug shows a request for gpu work, BOINC says, sorry you need to get rid of the other projects debt first. Just setting NNT for the other projects doesn't stop BOINC from having to finish the other projects work you have loaded. Unless I misunderstood your post that you set NNT AND finished all other projects work. Seti has the longest deadlines of any project that I'm aware of. Any other project with shorter deadlines will jump to the head of the line to be finished first.

As Richard constantly reminds us, BOINC is pretty darn smart about how it balances many projects simultaneously. Whether it works the way we want it to, is another story.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Nonono.

That line was from another project, and I only added it as an example of what was shown when boinc does not ask for work.

Here it is in its completenes.

yoyo@home 11/7/2018 09:45:27 PM Not requesting tasks: don't need (CPU: job cache full; AMD/ATI GPU: not highest priority project)

Seti looks like this:

SETI@home 11/7/2018 09:44:58 PM Requesting new tasks for AMD/ATI GPU

My PC is requesting tasks!

Just not given any!

---

Richard also mentioned that when running multiple projects it is best to run a pretty small cache in order for BOINC to handle the Resource Shares better between projects.

A few questions:
What resource share are you giving each project? (The scale is 0-1000 for each project)
What cache settings are you using? (There are two numbers, and their use is "not obvious")
Are any of the projects getting work reliably, if so, which?

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

As my settings are right now I have Seti at 160 resource share.

Other projects are at 1 or 2 in resource share. They are basically only backups. But do occasionally run a few wu's.

My cache is set at 0.75 / 2.25.

The other projects are getting work. Right now its mostly Collats, as its the only backup project I have allowed to use the GPU.

At the moment I have 5 backup projects, adding up to 8 in resource share (3 at 2 and 2 at 1). So about 1/20 of the resource share.

Collats has a 2 points resource share, so 1/80 of the GPU resource share. Shouldnt prevent Seti from getting work for a whole week.

My number are actually carefully selected, to get the resource shares I want.

When I check the scheduling priority for Seti vs Collats i get this.

Seti:

Scheduling priority -0.78

Collats:

Scheduling priority -235.23

If I understand these numbers, Seti has much higher priority right now.

I have been running BOINC for many many years, and understand the inner workings of the scheduler to some extend.

Something is wrong that is preventing my host from picking up work. The work is being requested from the server, just not handed out.

I'm of to bed, will see when I have the time to look into this again.

Work is being done afterall, so its no disaster.

---

Richard's standard answer for those running multiple projects is to reduce the cache size. Remember cache size is global across all projects. So 1 day of cache for one project may net you one work unit and 10,000 work units for another project. BOINC is still overcommitted on the backup projects compared to Seti. Easiest fix is to half the cache size. And everyone confuses the BOINC definitions on what the "store at least" days and "store up to an additional" days of work means. The primary determiner of the cache size is the first setting. The additional one is a provision from long past days when we were using dial-up communications and may not be able to contact the servers for a week at a time. With modern network comms which are always on, this setting is pretty much pointless. It also can prevent BOINC from contacting the servers to ask for more work. If you want the host to contact the servers at every 5 minutes scheduling connection, set the "additional days" setting to 0.01 days.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Resource share work on a long-term "debit" system - but can be upset when a project persists on dumping large numbers of tasks onto a user and those tasks have very short deadlines. One will frequently find a project will run solidly for an extended period of time, then get "kicked to the back" while another project takes over. This can takes or even weeks to sort out so things are running as you want them. This is the normal behaviour, but may not be what you want, or expect. One solution is to give your main project a share of 1000, then scale the others below that, and give your "real backup" projects a share of zero (this means "only get work when there's none available from my other projects", not "don't get any work").

And I would underline what Keith says about setting cache sizes when trying to run multiple projects - I would try a one or two day cache (the "store at least" part) and a very small refresh (the "store an additional" part), say 0.01days. Having the second figure low will trigger frequent calls for work, having it large means you might only call for work at long intervals (days....)

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Thanks for all the suggestions.

I'll try to reduce my cache size.

I still think the problem is elsewhere though.

My computer clearly asks the seti server for work, but gets the reply that no work is available.
Which is untrue, as there are several 100 thousands of WU's available.

21:36:44 | SETI@home | Requesting new tasks for AMD/ATI GPU
21:36:44 | SETI@home | [sched_op] AMD/ATI GPU work request: 346244.03 seconds; 1.00 devices
21:36:46 | SETI@home | No tasks are available for SETI@home v8

The servers dont repond that I have reached some limit (e.g. the 100 tasks limit), just that no work is avaiable for setiathome v8.

The servers dont know the internal debit of my boinc client. It simply responds to the request from my client. If my client asks for 50.000 seconds of work, the server tries to answer that request based upon previous numbers for crunch time and of course given the 100WU restriction and availability of work.

For some reason that doesn't work. I would like to know the reason why.

---

BOINC caches are across all projects running, so if you have the GPU cache full of tasks from another project you won't get any GPU tasks from SETI

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Well something changed on or shortly after Nov 1 on your computer, or another project. That is when you stopped getting work.
So what was changed, driver, project added, resources changed??

It is NOT the seti server acting up or this forum would be full of messages from other users with the same problem as you.

I went hunting for this Messege
Try enabling "cpu_sched_debug"

Thinking about high priority ... also if you have CPU tasks running High Priority and Seti reserving 1 thread for the GPU processing - the tasks won't go since the CPU is busy with it's backlog.

BoincTasks is a great program for debugging priority issues. You just can't miss the bright red highlight it puts on the task(s).
It is a wonderful all around program, and a must for multi computer users.

Thanks for the answers.

I have reduced the cache.

I'll let it run for some time, and see if job gets downloaded at one point.

I already use Boinctasks, and have done so for many years. I find it better than boinc manager in many ways, and it lets me monitor more computers easily.

Edit:

I just found the error:

A few days ago I changed the app that I crunch Setiv8 with to an optimized version for my Ryzen CPU. The app was mentioned in another thread.

In editing app_info.xml I accidentally put in a extra pair of
<app_info>
</app_info>
around the newly edited field, with the file names for the new app.

This of course confused boinc (it didn't show any errors though), and made it think that the file ended after the first </app_info>, after which all my GPU setiv8 apps where.

I removed these extra lines, restarted boinc, and work was sent to my PC immediatly.

Previously the servers saw that I didn't have valid programs for my request and acted accordingly.

I was basically only asking for AP work for my GPU, and that is hard to come by.

This was all my own fault, but it was a very hard error to find...

---

Aww, it is good that you fond it after much frustration.

Good that you figured it out. Basic troubleshooting. Stable system >> make change >> change expected results.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)