Well!!

I did it once again, this sucks but it's my own fault.

The common rule when upgrading to other boinc versions when you've got a cache full is to:

(0) RTFM about required posts in app_inf file if you run modified.

1. Disable Network access
2. Stop boinc & processes
3. Backup the Data directory just incase.
4. Uninstall old version
5. Install new version.
6. Watch what happens.

I ignored those steps and uninstalled my old version just to find out that i got a response that says:
"[SETI@home] [error] No app version for result: windows_intelx86 608 cuda"

And it wipes out my entire cache including .exe files and .dll files.
Leaving 2200 wu's uncredited and wiping the other 1300 in the cache..

Sorry once again wingmens.. Just a reminder what happens if you don't take precautions like i just didn't do. :(

Kind regards Vyper

---

_________________________________________________________________________
Addicted to SETI crunching!
Founder of GPU Users Group

Sorry to hear that!
I have upgraded to the most recent version a couple of days ago and everything went smoothly, quickly and without problems.
The only thing I did was to click on the new version button on the main page link.
For me, it worked.

---

...everything went smoothly, quickly and without problems...
For me, it worked.

It is because You are not using CUDA.

Windows versions of BOINC Manager 6.6.34 and 6.6.35 are known as bugged at CUDA-processing, not with CPU one.

My mistake was that i was curious about the newest unofficial version 6.6.35..
If i had downloaded the official version 6.6.31 it worked...

So this warning is a reminder of what is needed if you don't to have unforseen events like me :/ ..

Well upgrade to 6.6.35 = Wiped cache if not specifying other parameters in app_info which i can't find even if googling.

Upgrade to 6.6.31 = Works as nothing ever happened.

Gonna see how it works out with 6.6.31 with the pause/resume issue that i've been having when running large caches and returns aprox 1600 MB's / day.
Newer Boinc versions simply can't handle that efficiently and thinks that even if i have 4500 MB's in my cache it believes that i'm gonna miss deadline in two weeks and pauses ongoing results and starts new ones.
Even if my cache is emptied completely within 2,9 days acording to my math it stops wu's starting new ones, and that's ok with cpu tasks more than cuda tasks because it needs to preinitialise the data once again and wastes alot of cpu/gpu cycles.

Once i had around 200 paused wu's in my list , that looked funny :)

The version that copes so far good for me is good old 6.4.7 .
It hasn't behaved like that and this was the main reason for me that i wanted to test the new version ..

Kind regards Vyper

---

_________________________________________________________________________
Addicted to SETI crunching!
Founder of GPU Users Group

...everything went smoothly, quickly and without problems...
For me, it worked.

It is because You are not using CUDA.

Windows versions of BOINC Manager 6.6.34 and 6.6.35 are known as bugged at CUDA-processing, not with CPU one.

Здравствуйте!/Hello!
The only thing I have to mention about this is that for the last days, the credits are not being noted on the BOINC Statistics of the world page.
Usually they refresh daily the results. But I think it has nothing to do with the upgrade, since the missing credits issue is before the upgrade.

---

My mistake was that i was curious about the newest unofficial version 6.6.35..
If i had downloaded the official version 6.6.31 it worked...

So this warning is a reminder of what is needed if you don't to have unforseen events like me :/ ..

Well upgrade to 6.6.35 = Wiped cache if not specifying other parameters in app_info which i can't find even if googling.

Upgrade to 6.6.31 = Works as nothing ever happened.

Gonna see how it works out with 6.6.31 with the pause/resume issue that i've been having when running large caches and returns aprox 1600 MB's / day.
Newer Boinc versions simply can't handle that efficiently and thinks that even if i have 4500 MB's in my cache it believes that i'm gonna miss deadline in two weeks and pauses ongoing results and starts new ones.
Even if my cache is emptied completely within 2,9 days acording to my math it stops wu's starting new ones, and that's ok with cpu tasks more than cuda tasks because it needs to preinitialise the data once again and wastes alot of cpu/gpu cycles.

Once i had around 200 paused wu's in my list , that looked funny :)

The version that copes so far good for me is good old 6.4.7 .
It hasn't behaved like that and this was the main reason for me that i wanted to test the new version ..

Kind regards Vyper

Sorry to hear of your problems.

You can go up to 6.6.33. From what I could see 33 appears the same as 31 with the addition of some more debugging messages. Neither 31 nor 33 will shut down science apps on exit. Apparently it will if you run a service install. So depending on how you have it installed you may need to click on Advanced -> Shutdown connected client first or not.

---

BOINC blog

I've had all of those problems with both 6.6.31 and 6.6.33. 12 tasks were "waiting to run", and this system crashed out last night.

The odd part is Vyper is NOT running any tasks on the CPU (is that correct?), and I thought this problem only appeared in the CPU+GPU situation.

There is no good pattern to the problem yet.

Bob

What Vyper is seeing is a very specific bug, introduced for the first time with v6.6.34, and still present, by mistake, in v6.6.35

No earlier versions were affected by this bug - though there may well still be other bugs yet to be tracked down.

I have been given access to a private Windows_32 build of v6.6.36 (we're testing some other new features for another project), and that seems to be safe for CUDA use. So when v6.6.36 appears for Berkeley download, it'll be worth a try (backup, network off - usual precautions apply).

Could I suggest that anyone interested in testing the newest versions, before they're even announced by Rom Walton, subscribes to the boinc-alpha testing and reporting mailing list? That way, only one of us has to trash a perfectly good cache - I did mine last Thursday!

v6.6.36 is up and available in the place you got the earlier ones from - and installing it left my CUDA cache alone. Yay!

Out of idle curiosity, can't doing a detach/attach release all of those WUs back into the wild instead of waiting for them to time out?

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

Out of idle curiosity, can't doing a detach/attach release all of those WUs back into the wild instead of waiting for them to time out?

Yes, it can and does.

Make sure you preserve the residual contents of the seti project folder, after updating the final tasks from a flushed cache and before detaching: the whole folder gets deleted. If you've got a bespoke app_info.xml in there, and lots of imported applications, it saves time reassembling the collection.

v6.6.36 is up and available in the place you got the earlier ones from

It went straight to recommended version for all platforms, without so much as testing it. Fingers crossed.

---

Well!!

I did it once again, this sucks but it's my own fault.

The common rule when upgrading to other boinc versions when you've got a cache full is to:

(0) RTFM about required posts in app_inf file if you run modified.

1. Disable Network access
2. Stop boinc & processes
3. Backup the Data directory just incase.
4. Uninstall old version
5. Install new version.
6. Watch what happens.

I ignored those steps and uninstalled my old version just to find out that i got a response that says:
"[SETI@home] [error] No app version for result: windows_intelx86 608 cuda"

And it wipes out my entire cache including .exe files and .dll files.
Leaving 2200 wu's uncredited and wiping the other 1300 in the cache..

Sorry once again wingmens.. Just a reminder what happens if you don't take precautions like i just didn't do. :(

Kind regards Vyper

Just to add to this timely warning, I have been experiencing odd intermittent display shifts on my Vista64 machine so decided to reinstall the graphics driver last night - latest version 185.85.

Completely forgot in haste to lock down BOINC and somewhere through the process the CUDA apps disappeared along with all my CUDA tasks.

Apologies to anyone waiting on those tasks.

Doubly annoying as there was a scarcity of tasks being sent out last night.

Rebranded a few CPU tasks to GPU just to give it something to munch on.

All I can say is Doh! and of course next time I'll remember to disable everything before installing new graphic drivers.


John.

---

GPU Users Group