How many test WUs are generated?

I have machines here that do not get work, Windows32, Windows64 and Linux64. Only get work on Linux32.

Thank you.

---

How many test WUs are generated?

As with everything else having to do with computers, the answer is: It Depends.

The number of WU's we get depends on what's being tested.

If the developers are busy coding, we might not get any for a while. If they're testing a new feature, we'll get enough to test, then it will die off again. Then we'll get a few more, then nothing for a while.

However, if they're stress testing before a release, or trying to figure out a particular bug, then there are plenty to go around.

This project is for patient people who prefer quality over quantity. They need to be willing to go to the web site and examine the Results they return, and post any problems on the approriate thread.

Not hard, but not the standard "crunch crunch crunch" scenario either.

==Mike

---

Don't believe everything you think.

You can get a feel for it by checking the WUs in progress and successes last 24hrs on the homepage from day to day. Seems to be a couple thousand, at least, per release.

---

You can get a feel for it by checking the WUs in progress and successes last 24hrs on the homepage from day to day.

Good point...

==Mike

---

Don't believe everything you think.

Sometimes machines don't get work because their resource share is set low enough that BOINC is not requesting work. But as time passes and other projects are doing work, the debt to Ralph increases and you probably get work the next time.

---

How many test WUs are generated?

As with everything else having to do with computers, the answer is: It Depends.

The number of WU's we get depends on what's being tested.

If the developers are busy coding, we might not get any for a while. If they're testing a new feature, we'll get enough to test, then it will die off again. Then we'll get a few more, then nothing for a while.

However, if they're stress testing before a release, or trying to figure out a particular bug, then there are plenty to go around.

This project is for patient people who prefer quality over quantity. They need to be willing to go to the web site and examine the Results they return, and post any problems on the approriate thread.

Not hard, but not the standard "crunch crunch crunch" scenario either.

==Mike

I guess my question really was how many test WUs are generated for W32, W64, L32 and L64 on each group of test WUs generated.

Are test WUs only created for one flavor of OS on occasion?

---

I believe that WUs are created without any specific designation as to platform. In fact I'm pretty sure I've seen WUs where 1st person was Windows and second was Linux or Mac. So it is just a matter of who the task gets assigned to.

---

I believe that WUs are created without any specific designation as to platform. In fact I'm pretty sure I've seen WUs where 1st person was Windows and second was Linux or Mac. So it is just a matter of who the task gets assigned to.

Ahhh, no HR.......guess that makes it hard for Linux puters to get Linux WUs.

---

I believe that WUs are created without any specific designation as to platform. In fact I'm pretty sure I've seen WUs where 1st person was Windows and second was Linux or Mac. So it is just a matter of who the task gets assigned to.

Ahhh, no HR.......guess that makes it hard for Linux puters to get Linux WUs.

...umm, actually... that's the only type that a Linux puter could get. Just means they would have to burn through a lot of Windows task requests to land lots of Linux requests. Everyone has that random chance of requesting work when it is available... Linux is just outnumbered.

---

I believe that WUs are created without any specific designation as to platform. In fact I'm pretty sure I've seen WUs where 1st person was Windows and second was Linux or Mac. So it is just a matter of who the task gets assigned to.

Ahhh, no HR.......guess that makes it hard for Linux puters to get Linux WUs.

...umm, actually... that's the only type that a Linux puter could get. Just means they would have to burn through a lot of Windows task requests to land lots of Linux requests. Everyone has that random chance of requesting work when it is available... Linux is just outnumbered.

I obviously don't understand the process of generating WUs. If they say they "need" more Linux WUs processed, then doesn't it stand to reason they need to generate more Linux WUs. Why is that linked to Windows WUs?

---

I obviously don't understand the process of generating WUs. If they say they "need" more Linux WUs processed, then doesn't it stand to reason they need to generate more Linux WUs. Why is that linked to Windows WUs?

It means they either need more linux hosts to attach and process wu's, or existing linux hosts to increase their resource share and process more wu's. There are no linux or windows only wu's.

---

When a work unit is created, it can be run anywhere. If a Linux host requests the work, then it runs on Linux. If more Linux hosts attach to Ralph, or they have a higher resource share allocated to Ralph, then, on average, you will have more Ralph work being tested on Linux.

---

How do we get the Windows computers to shut down for a day or two to test Linux WUs?

---

The randomness of the work generation is partly the problem as well as the lower number of Linux machines.
I have had 3 Linux computers attached 24/7 but only get a random WU now and then.
The last time one of them got a WU was on the 20/12/07, yet it tries every day.
It would appear that the Windows machines are requesting the work faster than Linux hence we don't get the work.

As for Windows I have one of those and it has 1 WU since the 18/12/07.

I am trying to grab the work before those marauding Windows machines do but I am just outnumbered.

I did pick up the problem that was eventually fixed (and reported it), but the project did not react to just the one report as they needed more proof it was a wider issue.

If I had more money then I could add more computers.

---

The randomness of the work generation is partly the problem as well as the lower number of Linux machines.
I have had 3 Linux computers attached 24/7 but only get a random WU now and then.
The last time one of them got a WU was on the 20/12/07, yet it tries every day.
It would appear that the Windows machines are requesting the work faster than Linux hence we don't get the work.

As for Windows I have one of those and it has 1 WU since the 18/12/07.

I am trying to grab the work before those marauding Windows machines do but I am just outnumbered.

I did pick up the problem that was eventually fixed (and reported it), but the project did not react to just the one report as they needed more proof it was a wider issue.

If I had more money then I could add more computers.

I had more Linux computers trying to get work and decided it was pointless.

The project needs to use homogenous redundancy or some other method of assigning WUs specifically to Linux or whatever OS they are working with at the moment.

---

Y'all know how to twist a gal's arm. :-)

I just attached my Fedora 7 host.

It now has one in progress, but it's only been running for a few minutes.

It would be nice if work was marked for a specific OS. It would help the testing out immensely. I'm not sure if there is a way to do it outside HR though. And I don't know how HR would work with a quorum of 1.

---

Kathryn :o)
The BOINC FAQ Service
The Unofficial BOINC Wiki
The Trac System
More BOINC information than you can shake a stick of RAM at.

Y'all know how to twist a gal's arm. :-)

I just attached my Fedora 7 host.

It now has one in progress, but it's only been running for a few minutes.

It would be nice if work was marked for a specific OS. It would help the testing out immensely. I'm not sure if there is a way to do it outside HR though. And I don't know how HR would work with a quorum of 1.

If they had to go to quorum of 2, it would still benefit the project to be able to direct to Linux.

I put another Linux box on this morning and get nada.

---

Y'all know how to twist a gal's arm. :-)

I just attached my Fedora 7 host.

It now has one in progress, but it's only been running for a few minutes.

It would be nice if work was marked for a specific OS. It would help the testing out immensely. I'm not sure if there is a way to do it outside HR though. And I don't know how HR would work with a quorum of 1.

If they had to go to quorum of 2, it would still benefit the project to be able to direct to Linux.

I put another Linux box on this morning and get nada.

Big problem with quorum of two is that people can set their own run time. So a 2 hour unit won't validate against a 4 hour one.

---

Kathryn :o)
The BOINC FAQ Service
The Unofficial BOINC Wiki
The Trac System
More BOINC information than you can shake a stick of RAM at.

Y'all know how to twist a gal's arm. :-)

I just attached my Fedora 7 host.

It now has one in progress, but it's only been running for a few minutes.

It would be nice if work was marked for a specific OS. It would help the testing out immensely. I'm not sure if there is a way to do it outside HR though. And I don't know how HR would work with a quorum of 1.

If they had to go to quorum of 2, it would still benefit the project to be able to direct to Linux.

I put another Linux box on this morning and get nada.

Big problem with quorum of two is that people can set their own run time. So a 2 hour unit won't validate against a 4 hour one.

I did not know that........interesting in trying to think of all the permutations.

Maybe they can make two projects (like Malaria and others), where users opt in or out. Linux could be one project, Windows another, etc.

---

Y'all know how to twist a gal's arm. :-)

I just attached my Fedora 7 host.

It now has one in progress, but it's only been running for a few minutes.

It would be nice if work was marked for a specific OS. It would help the testing out immensely. I'm not sure if there is a way to do it outside HR though. And I don't know how HR would work with a quorum of 1.

It looks like they are working on that:

zo 06 jan 2008 04:54:20 CET|ralph@home|Sending scheduler request: To fetch work
zo 06 jan 2008 04:54:20 CET|ralph@home|Requesting 766 seconds of new work
zo 06 jan 2008 04:54:25 CET|ralph@home|Scheduler RPC succeeded [server version 509]
zo 06 jan 2008 04:54:25 CET|ralph@home|Message from server: No work sent
zo 06 jan 2008 04:54:25 CET|ralph@home|Message from server: (there was work for other platforms)
zo 06 jan 2008 04:54:25 CET|ralph@home|Deferring communication for 1 days 0 hr 0 min 0 sec
zo 06 jan 2008 04:54:25 CET|ralph@home|Reason: requested by project