When I run QOpt, I get 4 lines saying "INFO: No tasks running with the specified criteria".

I have never seen that message before. Are you sure that the Rescheduler is in the BOINC folder and not your SETI folder?

Well, there goes that theory that not many people are running CUDA tasks now with all the GUPPI work that is being delivered. I sort of assumed most everyone would be running the SoG app for the predominance of GUPPI's we now have. And not many people are running Petri's special CUDA sauce since it is still considered beta. So I wondered whether the apps didn't have much exposure to the CUDA apps and might have an obscure issue.

---

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

A proud member of the OFA (Old Farts Association)

Ok - Thanks to all for your help.

I've got the rescheduler installed in the same directory as the the client_state file, I've QOpt downloaded, running and talking to GUPPIRescheduler and I've updated BOINC to the latest version.

I have 20 "normal" WU's on the CPU's and 40 odd Guppis on the GPU's

When I run QOpt, I get 4 lines saying "INFO: No tasks running with the specified criteria".
Apart from from that it all seems to go through the motions of rescheduling the WU's quite happily but nothing happens. I still have guppis on the GPU's while having "normal" WU's on CPU's.

QOpt.log shows no errors, the problem seems to be with the Rescheduler. Does it work on XP ? Or should I be looking at something else ?

Again TIA

T.A.

. . Hi TA

. . I am sorry but I no longer have XP running on any rigs that are crunching. The only thing I might suggest is to install Lunatics and then you will know that the aps crunching are within the scope of the rescheduler. If it still fails to identify and move them then the problem may be an incompatibility with XP. The rescheduler relies on first identifying Arecibo work in the CPU queue that can be moved and will only move as many guppi tasks from the GPU queue as it has moved normals from the CPU queue. So failing to identify the first group will result in no action. But the more common error message is that it has failed to identify the version of tasks that are running. Maybe that will steer you towards a solution.

Stephen

.

I looked back through the entire thread looking for anyone that is running CUDA successfully with the rescheduler. Nobody else in the thread has explicitly said they have the program running CUDA tasks. It would be helpful is anyone reading this thread and using the program with CUDA to say what luck they are having with it.

. . Hi Keith

. . I am running CUDA50 on Mi_Burrito (Core2 Duo, GT730, Win10 and the very latest BOINC{7.6.33} and previously 7.6.22) and using the qopt-rescheduler successfully on that rig. I do know that if you are running stock and have mixed apps running then the rescheduler can fail to run because it does not know which app to use for them. But as I said to TA, that usually gives the "failed to identify app version running" message. I recall there was a message at one point that stipulated running Lunatics to keep the apps homogenous. But never having received the message he cites I cannot say that is the problem. I do believe though that that message is from qopt not from the rescehduler itself [edit: Jimbo says no!]. I think reading through the help file with qopt as in "qopt /h" might reveal a solution.

Stephen

.

Well, there goes that theory that not many people are running CUDA tasks now with all the GUPPI work that is being delivered. I sort of assumed most everyone would be running the SoG app for the predominance of GUPPI's we now have. And not many people are running Petri's special CUDA sauce since it is still considered beta. So I wondered whether the apps didn't have much exposure to the CUDA apps and might have an obscure issue.

. . I think Petri's app is more Alpha then Beta, you have to compile it to run it and you need to know a lot to get it to work on your platform. I think it is still months away from hitting a Beta (Lunatics?) release. But I can assure you there is no issue with CUDA50.

Stephen

.

Stephen, I think I know where you were going with your thinking in your post. But Terror is not running stock apps controlled by client_state making supposedly smart,efficient app choices. He is running the Anonymous platform and unless he cooked up his app_info from scratch, probably used the Lunatics installer. That forces you into making one single MB GPU app choice so not the case where multiple GPU apps are confusing the rescheduler and it bails out.

---

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

A proud member of the OFA (Old Farts Association)

Stephen, I think I know where you were going with your thinking in your post. But Terror is not running stock apps controlled by client_state making supposedly smart,efficient app choices. He is running the Anonymous platform and unless he cooked up his app_info from scratch, probably used the Lunatics installer. That forces you into making one single MB GPU app choice so not the case where multiple GPU apps are confusing the rescheduler and it bails out.

. . Hi Keith,

. . You would think I would learn to actually look at the guys setup before giving advice :( Oops!

. . Well another perfectly good theory shot down in flames. So maybe the finger of blame lies in the directories that BOINC is installed in. But Jimbo made several corrections to overcome that issue so I doubt it. The only directory related issue that I know of is that the rescheduler MUST be in the BOINC directory and not the project directory or anywhere else. Funny how computing can always manage to confound us :)

Stephen

.

. . Well another perfectly good theory shot down in flames. So maybe the finger of blame lies in the directories that BOINC is installed in. But Jimbo made several corrections to overcome that issue so I doubt it. The only directory related issue that I know of is that the rescheduler MUST be in the BOINC directory and not the project directory or anywhere else. Funny how computing can always manage to confound us :)

Stephen

.

The log will tell the story ...

---

Anyone remember what the solution was to getting rid of the Red Ball and "reconnecting to client" timeout error?
Cliff Harding is having an issue with this since upgrade to Win 10.
(GR and QOpt not in play, happens to him when an excluded app is run and then exits).
For the life of me I can't recall if there was a solution needed or if it just went away on its own?
Or is it still happening and I just don't know it ...

---

I believe I remember the topic but I can't put my finger in what thread it came up in. I did a search on "reconnecting to client" but it produced results several years old. The gist of those comments were about either reinstalling BOINC or keeping AV programs fingers off the BOINC directories. That doesn't fit with what I remember the last conversations were about. I think the problem was explained by Richard. Hope somebody else with better memory or forum notes can set us straight.

---

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

A proud member of the OFA (Old Farts Association)

I believe I remember the topic but I can't put my finger in what thread it came up in. I did a search on "reconnecting to client" but it produced results several years old. The gist of those comments were about either reinstalling BOINC or keeping AV programs fingers off the BOINC directories. That doesn't fit with what I remember the last conversations were about. I think the problem was explained by Richard. Hope somebody else with better memory or forum notes can set us straight.

There were some comments about it back around pg 22-23 of this thread, but nothing definitive. When I released 1.02k I added the -cx option to add x seconds of delay to client restart, but don't believe that had a positive effect.
I had the issue here, but haven't seen it since early Nov, and don't remember why?
Frustrating ... I need to keep better notes in the unlikely event I ever tackle another thing like this.

---

I believe I remember the topic but I can't put my finger in what thread it came up in. I did a search on "reconnecting to client" but it produced results several years old. The gist of those comments were about either reinstalling BOINC or keeping AV programs fingers off the BOINC directories. That doesn't fit with what I remember the last conversations were about. I think the problem was explained by Richard. Hope somebody else with better memory or forum notes can set us straight.

. . If my memory serves me well Richard mentioned an issue about timing on the restart, saying that things like the rescheduler may be jumping the gun a little. A second attempt to start the manager/client seemed to work in the cases mentioned.

Stephen

.

. . If my memory serves me well Richard mentioned an issue about timing on the restart, saying that things like the rescheduler may be jumping the gun a little. A second attempt to start the manager/client seemed to work in the cases mentioned.

Stephen

.

That was the discussion, and that's where I added the -cx delay restart option to QOpt, but even if we ran it up to 15 seconds, it didn't help. 0,5,10 or 15 seconds didn't change anything.
Just seems weird that it has vanished without a trace as an issue.

---

Yea, I don't remember any definitive answer to the issue, but the topic did seem to vanish.

Basically it was to just restart BOINC wasn't it? I do remember mine doing it occasionally, and sometimes it did require a reboot. Tasks were running, but couldn't see them.

With Exclusives apps, I would think it might be in the BOINC upgrades, Have you upgraded Keith?

It shouldn't have anything to do with short duration timing for an Exclusive app since it is not a 5 second thing (normally).

I don't have the problem, Brent. Jim said it was Cliff Harding who was having issue restarting after an exclusive app was polled. I am on the latest 7.6.33 and haven't had any BOINC issues for at least a couple of revisions back. I haven't seen the problem being discussed.

---

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

A proud member of the OFA (Old Farts Association)

I don't have the problem, Brent. Jim said it was Cliff Harding who was having issue restarting after an exclusive app was polled. I am on the latest 7.6.33 and haven't had any BOINC issues for at least a couple of revisions back. I haven't seen the problem being discussed.

I'm on version 6.10.60 and there's been no problems for me, but then I don't move things about either.

Cheers.

I'm running Win10 Pro 64bit and it does take the manager longer to connect with the client than it does for Win7/Vista.
There have been one or 2 occasions where the red dot is there after 15 seconds or so, in which case I just exited BOINC, waited 5 seconds & restarted. It connects OK after that.

---

Grant
Darwin NT

I too have noticed the much longer time to connect with the client on Windows 10 64 bit. It's pretty much instantaneous on my Windows 7 64 bit machines.

---

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

A proud member of the OFA (Old Farts Association)

Yea, I don't remember any definitive answer to the issue, but the topic did seem to vanish.

Hate it when that happens :)

With Exclusives apps, I would think it might be in the BOINC upgrades, Have you upgraded Keith?
It shouldn't have anything to do with short duration timing for an Exclusive app since it is not a 5 second thing (normally).

It is Cliff Harding that was having the issue, but a good point to ensure the BOINC client is up to date ...

---

I'm running Win10 Pro 64bit and it does take the manager longer to connect with the client than it does for Win7/Vista.
There have been one or 2 occasions where the red dot is there after 15 seconds or so, in which case I just exited BOINC, waited 5 seconds & restarted. It connects OK after that.

Alternately, just double-click the BOINC tray icon, then File, Select computer and arrow key will fill the computer to connect fields. Saves on restarts.

---