What would cause one client to get this message and not be able to download work units when all of my other clients are just fine? All clients use the same preferences.

Thanks.

---

Rob,

If you installed BOINC into one machine, and then copied everything from that folder to another machine, in order to "install" boinc there, you would get this situation.

When boinc first connects to a project it gives that computer a unique number, which it stores in a file in the boinc folder.
Every time the server sends a WU to that machine (or any machine using the same number) it adds one to that host's daily WU download count.

If you had several hosts setup this way (with the same number), they would soon reach the total daily quota allowed for one host number.

Each machine should get its own number.

If you use regular install of boinc and then attach you will get a new number.

If you use regular boinc install, and only copy the file named "account_setiathome.berkeley.edu.xml" from the original machine to the others, they will get a new number.

> What would cause one client to get this message and not be able to download
> work units when all of my other clients are just fine? All clients use the
> same preferences.
>
> Thanks.
>

Since you're hiding your computers, and haven't supplied the computer-id to the one having problem, can only guess...

But a possible reason is this. A resent change to scheduling-server will decrease the computers daily quota if either reporting an error or doesn't return before the deadline. If you're only returning errors, you will eventually have a daily quota of 1.
Immediately when you reports a "success"-result the quota will increase again, but not above whatever the project have set as a quota.

> But a possible reason is this. A resent change to scheduling-server will
> decrease the computers daily quota if either reporting an error or doesn't
> return before the deadline. If you're only returning errors, you will
> eventually have a daily quota of 1.
> Immediately when you reports a "success"-result the quota will increase again,
> but not above whatever the project have set as a quota.

It would be nice to have a FAQ list as on the website of E@H.

>A resent change to scheduling-server will
> decrease the computers daily quota if either reporting an error or doesn't
> return before the deadline. If you're only returning errors, you will
> eventually have a daily quota of 1.
> Immediately when you reports a "success"-result the quota will increase again,
> but not above whatever the project have set as a quota.
>

I noticed this a week or two ago and I personally love it. I like to keep a 10 day cache of workunits. Unfortunately, the scheduler used to download 15 days of work...meaning I'd miss deadlines. Now, I can be assured of not going over anymore, yet still maintaining a huge cache. :)

Oh, and I know some people may flame me for "hoarding" workunits, but my reason for doing this is simply because I'm not interested enough in any of the other current BOINC projects to attach to them. Once Planetquest comes out, I'll be lowering my cache to about 5 days or so.

---

You will be assimilated...bunghole!

Once Planetquest comes
> out, I'll be lowering my cache to about 5 days or so.
>

Planetquest...what's that? I haven't heard of that one yet.

Torrey Lauer
Rainbow Sky Travel
www RainbowSkyTravel com

> Once Planetquest comes
> > out, I'll be lowering my cache to about 5 days or so.
> >
>
> Planetquest...what's that? I haven't heard of that one yet.
>
> Torrey Lauer
> Rainbow Sky Travel
> www RainbowSkyTravel com
>


http://www.planetquest.org/

---

You will be assimilated...bunghole!

> > Once Planetquest comes
> > > out, I'll be lowering my cache to about 5 days or so.
> > >
> >
> > Planetquest...what's that? I haven't heard of that one yet.
> >
> > Torrey Lauer
> > Rainbow Sky Travel
> > www RainbowSkyTravel com
> >
>
>
> http://www.planetquest.org/
>

Am I right in that I read that the project won't be live until 2006? Is that correct?

It sounds like a great project! I'm excited about it. Thanks for posting the link.

Torrey Lauer
Rainbow Sky Travel
www RainbowSkyTravel com

> Am I right in that I read that the project won't be live until 2006? Is that
> correct?
>
> It sounds like a great project! I'm excited about it. Thanks for posting the
> link.
>
> Torrey Lauer
> Rainbow Sky Travel
> www RainbowSkyTravel com

Old Planetquest Thread

It very well could have been error related. All of the machines I have were original installs BTW. This particular one did have some errors prior to the last power outage at Berkeley. I reset the project though. Yesterday it said that the daily quota was 1 and I figured that someone might be rationing work units. But today I noticed that none of the other machines had this problem. I'll let it go and see what happens.

I keep my cache settings at 8 days. They were at 5 but I just recently upped that to 8. I believe the clients should only dowload what it believe it can actually do in the period of time you specify. So if you say 5 days, it should be downloading only the number of WUs that the client can complete in 5 days. I imagine it derives that from the bench testing it does from time to time. This is my observation anyway. It would be interesting to find out why it was falling behind and not returning on time.

As for hiding the machines, yes but its out of necessity. Most of them are on a domain and S@H reports not only the machine name, but the fully qualified name on the domain. Sometimes this contains partial user names and I don't want any of that available to the public.

Thanks for the info. I'll watch it for a while.


Rob

---

> As for hiding the machines, yes but its out of necessity. Most of them are on
> a domain and S@H reports not only the machine name, but the fully qualified
> name on the domain. Sometimes this contains partial user names and I don't
> want any of that available to the public.
>
only you can see the machine names and IPs, we would just get host #s

check out anyone's list to see this (mine are not hidden)

happy crunching!

Micah

---

S@h Berkeley's Staff Friends Club

>I keep my cache settings at 8 days. They were at 5 but I just recently upped that
>to 8. I believe the clients should only dowload what it believe it can actually
>do in the period of time you specify. So if you say 5 days, it should be
>downloading only the number of WUs that the client can complete in 5 days. I
>imagine it derives that from the bench testing it does from time to time. This is
>my observation anyway. It would be interesting to find out why it was falling
>behind and not returning on time.

If you are using 4.19...this is not the case, the client downloads twice what your cache setting is. 4.25 downloads what your cache setting is. If you have 4.19 and set your cache to 8 days you will download 16 days worth of work...2 days over the deadline.

Don

---

>
> If you are using 4.19...this is not the case, the client downloads twice what
> your cache setting is. 4.25 downloads what your cache setting is. If you have
> 4.19 and set your cache to 8 days you will download 16 days worth of work...2
> days over the deadline.
>

Another resent change to the scheduling-server ensures you'll not download work that has expected run-time past the deadline.

But, there is still possible will miss the deadline for a couple reasons...
If you afterwards attach to more projects, or decreases resource_share on a project, the run-time will increase so can pass the deadline.
If your actual run-time is longer than the expected based on your benchmark, you can still pass the deadline. This is mostly a problem for Einstein@home, but atleast for Celerons it seems to be a problem.
If you're completely out of work for a project, you'll download 1 wu regardless of fast enough to return before deadline or not.
A couple projects haven't upgraded their scheduling-servers yet, so can hand out much more work than can manage to return.