Ah OK. In that case I think you will get the new names once BOINC has contacted the WCG scheduler and gotten the new naming convention in. Do you (or anyone else) still have one machine that shows the names in the old format, so you can test if Boinc automatically updates?

Yes, I have a few hosts showing the old format names and they don't seem to be updating them on their own or with manual updates. The hosts have been running 5.8.0 since before Christmas and 5.8.1 for the last 48 hours with a 'connect to' period of 0.2 days, so they've contacted the WGC scheduler a few times now.

For a test, run one of them on No New Tasks, then when the cache is used up/reported, reset that project. Allow new tasks after you did the reset. Make note if it downloads a new application.

Then report that back to me and I'll ask the devs.

---

For a test, run one of them on No New Tasks, then when the cache is used up/reported, reset that project. Allow new tasks after you did the reset. Make note if it downloads a new application.

Then report that back to me and I'll ask the devs.

I've done that, and the client then gets the updated user friendly application names and re-downloads the applications. That's what I meant in my original post about resetting the project.

Maybe the client only updates the user friendly application name when it downloads a new application. I haven't been running 5.8 long enough to see a project release an updated application to see if that does it. If this is what's happening at least it explains the differences I'm seeing between hosts.

---

It would also make it a WCG problem. If their scheduler doesn't tell a new application or changed application is available, BOINC won't download it automatically.

But I'll email Rom and ask anyway.

---

Thanks, it's only a cosmetic issue at the end of the day though.

My biggest problem with BOINC at the moment are the repeated crashes of the core client that many people have been seeing since early November (see my "BOINC keeps crashing" thread in the core client forum). The crashes are still happening with 5.8.1.

---

Thanks, it's only a cosmetic issue at the end of the day though.

My biggest problem with BOINC at the moment are the repeated crashes of the core client that many people have been seeing since early November (see my "BOINC keeps crashing" thread in the core client forum). The crashes are still happening with 5.8.1.

I saw your post in the other thread and already informed David and Rom about it.
Although I tested Rebirther's procedure and can't get mine to crash.

---

Thx Ageless, Iam still using WinXP SP1 with all updates, latest directx 9c. But Marky-UK using SP2 with same result.

Shouldn't an update allow the user friendly names to be updated? Bug?

Um, explain. What are user friendly names? Give an example or two.

WGC has an application called "fcg1", BOINC Manager uses that name in the Task list in the Advanced View. However, in the Simple View on the "fgc1" tab it shows the Application as "Genome Comparison" instead - the "Genome Comparison" is the user_friendly_name that comes from the project.

BoincView also shows the Application as "Genome Comparison" instead of "fgc1".

That's on a host that's working and its what I'd expect to see on all hosts running 5.8.1 and WGC.

On a different host running 5.8.1, the application is still showing as "fgc1". If I reset the WGC project on that host, it'll start showing "Genome Comparison" instead, but I think just an Update should do it.

WGC uses friendly names for all its projects, hdc is Help Defeat Cancer, faah becomes FightingAIDS I think. Rosetta uses them too, but it just uses "Rosetta" as the friendly name for "rosetta" (different case).

Answer from David Anderson:

I checked in a change to use user-friendly names
-- David

---

Answer from David Anderson:

I checked in a change to use user-friendly names
-- David

Just tested that in 5.8.2 and its changed the application name shown in the BOINC Manager Task list.

It hasn't changed anything else though, but then if the client only gets the user-friendly name when it downloads an application I guess there's not much else that can be done. I'll have to wait until a new project application is released or reset the projects to get the names updated.

The messages from Rom about the libCurl crashes are encouraging though. :-)

---

5.8.x has to learn again from the beginning. If you are attached to 3 projects, a 5.4.x version would get you 1 day worth of work combined all over the projects. Now depending on resource share, the debts for the various projects and if you want to give Boinc time to learn again, it will fetch that day worth of work per project. Within boundaries of course.

Jord,

You wrote the above in response to this post from Bob Guy. I'm not sure I understand exactly what you mean by the statement. You seem to be saying that 5.4.11 would only download a fraction of the cache size according to the projects resource share whilst 5.8.x will download the full cache size from each project. As a simple example, if I had three equally resourced projects and a 1 day cache size, 5.4.11 would end up with 3 x 0.333 = 1 days work whilst 5.8.x would end up with 3 x 1 = 3 days work???? Is this what you are saying or am I misreading it? I ask because this seems to be contrary to my own experience with 5.4.11, particularly where you make a change to the cache size, say 1 day going to 2 days and a low resourced project (eg 5%) gets an unwanted full 2 days of work as if it were on 100%.

The reason I'm chiming in here is that I've just done a few 5.8.x installs and I'm seeing exactly what Bob Guy reported here. The relevant bit from his message states:-

The 'time to completion' seems to be increasing instead of decreasing. I have WUs with to complete times of 00:37:xx that complete at about 00:21:xx but the 'time to completion' of the queued WUs continues to increase. Shouldn't they decrease?

Under 5.4.11 when the actual crunch time is less than the estimate, BOINC reduces the DCF so that the new estimate will reduce by 10% of the difference between the old estimate and the actual crunch time. Using Bob Guy's example of a 37min estimate and a 21min actual crunch time, the downward adjustment to DCF should reduce the new estimate from 37 to 35.4min ie 1.6min reduction so that the two numbers will converge.

I can confirm a very similar behaviour to that reported by Bob Guy. I have observed a number of results where the actual crunch time is less than the estimate, yet 5.8.x continues to increase the DCF to make the gap widen rather than narrow. Initially I thought this might be a runaway process but after further observation the estimate reaches a plateau and doesn't increase any further.

Obviously there must be a reason for this - probably along the lines of "Tough, that's just the way things work now" :). Unfortunately, I kinda like the old way where DCF did a marvellous job of bringing the estimated time and the actual time into good alignment :).

Do you have any knowledge of the reasoning behind this new observed behaviour?

---

Cheers,
Gary.

If you set the queue to 0.5 days, that will be divided among the potentially runnable projects by resource share. It ended up with a fairly complicated set of restrictions.

Examples:

Project    RShare    LTD    contactable?    project queue.
A          100        0        yes          0.25 days
B          100        0        yes          0.25 days
_________________________________________________
A          100        50000     yes         0.5 days
B          100        -50000    yes         0 days.
________________________________________________
A          100        0         yes         0.5 days
B          100        0         no          0 days
________________________________________________
A          90         0         yes         0.45 days
B          10         0         yes         0.05 days
________________________________________________
A          100        50000     no          0 days
B          100        -50000    yes         0.5 - total time remaining on all WUs.

Projects have to be contactable (no pending communications)
Not suspended or NNW.
Have an LTD > -project switch interval.

If there is a result in EDF, then the queue will only fill if the amount of work remaining is < connect interval.

Projects with a result in EDF will not get work in any case.

Projects with low LTD will get work if there is no better project to get work from and the total amount of work is < connect interval.

Otherwise, it will divide the connect interval evenly.

JM7 fixed the bug that attempted to download 10 days of work for all 10 projects attached (100 days worth of work with the typical deadline being about 2 weeks - not a good idea).

So in answer to your question, Gary, I misread it. :-)

Now about those increasing to completion times, I see that effect only when I switch applications (Primegrid) or when the project has results with different runtimes. The DCF is per project, not per result or application.

---

Iam missing now the good old red font color of "no work" and "unrecoverable error" status messages in 5.8.x

Answer from David Anderson:

I checked in a change to use user-friendly names
-- David

I think this change is a little broken and introduces a couple of bugs when BOINC Manager 5.8.2 is used to connect to a remote BOINC Client 5.4.11:

- In Advanced View, the Tasks show no Application at all now, just the version number.

- In Simple View, the BOINC Manager shows no tasks at all! It just says "Error No work available to process".


It seems to me like the application name should fall back to the old application name if the user-friendly name is missing.

I'm not 100% sure if the Simple View isn't working because of this change or something else that changed in 5.8.2, but it worked with BM 5.8.1 connecting to a remote 5.4.11.

---

Forwarded to the powers that be.

---

I am having some problems with 5.8.2 WU, WU, WU Client State says "Success", but no credit???

Question is, what happened to the Windows download of 5.8.1, I had downloaded that and it was working fine...but now, a windows 5.8.1 is "NOT" available for download.....any ideas???

If this keeps up, I will go back to 5.2.13tx36

I am having some problems with 5.8.2 WU, WU, WU Client State says "Success", but no credit???

pls ask the project admin there.

Question is, what happened to the Windows download of 5.8.1, I had downloaded that and it was working fine...but now, a windows 5.8.1 is "NOT" available for download.....any ideas???

If this keeps up, I will go back to 5.2.13tx36

You will find all versions here: http://boinc.berkeley.edu/dl/?C=M;O=D

Question is, what happened to the Windows download of 5.8.1, I had downloaded that and it was working fine...but now, a windows 5.8.1 is "NOT" available for download.....any ideas???

5.8.1 still had big bugs in it. So the new download available now is 5.8.2, the latest TEST version. When this one comes up clean, as in not that many show stopping bugs, it could well become the recommended version.

If this keeps up, I will go back to 5.2.13tx36

"They release too many (test) versions, I can't take it anymore, I am going back to my old version!"

I don't get it. Don't you want BOINC to become better?

---

Rebirther..Appreciate the info on where the complete index of apps are located..

and to Ageless ,yes, I am always looking forward to better programs being released by BOINC...

I posted my remarks here and also in the project..admin said they were reconfiguring the WU's that I referred to..and that was the reason for zero credit...

Just thought i would mention that I have been successfully using the "TEST"
5.8.1 and was also willing to try the "TEST" 5.8.2...just reporting what I had...

If you set the queue to 0.5 days, that will be divided among the potentially runnable projects by resource share. It ended up with a fairly complicated set of restrictions.

Examples:

....

Thanks very much for those examples. Very useful to have that information.

JM7 fixed the bug that attempted to download 10 days of work for all 10 projects attached (100 days worth of work with the typical deadline being about 2 weeks - not a good idea).

Glad to hear that behaviour has been fixed.

Now about those increasing to completion times, I see that effect only when I switch applications (Primegrid) or when the project has results with different runtimes. The DCF is per project, not per result or application.

Yes, the DCF is per project, which effectively means per app as the projects I crunch have a single app. I'm observing this growth in DCF which is entrenching a permanent and quite large error between actual crunch time and estimated crunch time on about six boxes that I've upgraded to 5.8.x. A typical example is a box with an actual crunch time of about 2 hrs 10 mins which has all cached work listed at around the 3 hour mark. About 2 days ago I had manually adjusted the DCF to give an estimate of 2 hrs 10 mins and 5.8.x has kindly undone my manual help and adjusted the value upwards again :). I'm hoping someone will chime in who can explain why DCF is no longer working as it used to in 5.4.11.

---

Cheers,
Gary.

Iam missing now the good old red font color of "no work" and "unrecoverable error" status messages in 5.8.x

Not sure if errors are no longer in red, as I don't see them that many.

But informational messages such as No work from project don't need to be in red, as they aren't errors. They even have their own line now.
e.g. 12/01/2007 13:56:19|Pirates@Home|Deferring communication for 1 minutes and 0 seconds
12/01/2007 13:56:19|Pirates@Home|Reason: requested by project
12/01/2007 13:56:19|Pirates@Home|Deferring communication for 2 minutes and 13 seconds
12/01/2007 13:56:19|Pirates@Home|Reason: no work from project

---

Answer from David Anderson:

I checked in a change to use user-friendly names
-- David

I think this change is a little broken and introduces a couple of bugs when BOINC Manager 5.8.2 is used to connect to a remote BOINC Client 5.4.11:

- In Advanced View, the Tasks show no Application at all now, just the version number.

- In Simple View, the BOINC Manager shows no tasks at all! It just says "Error No work available to process".


It seems to me like the application name should fall back to the old application name if the user-friendly name is missing.

I'm not 100% sure if the Simple View isn't working because of this change or something else that changed in 5.8.2, but it worked with BM 5.8.1 connecting to a remote 5.4.11.

David told me he fixed this now as well. So it's available in the next version of BOINC, or the release version of 5.8.2 if that one is the RC. :)

---