As I said, the Science App uses the BOINC libraries from bonic-master. The App will not compile without those BOINC libraries. The BOINC API is from those BOINC Libraries, Not the Science App.
libboinc_api.a : /Users/Tom/boinc/api

As I said, the Science App uses the BOINC libraries from bonic-master. The App will not compile without those BOINC libraries. The BOINC API is from those BOINC Libraries, Not the Science App.

Correct. That's how it's made.

But it lives and runs inside the application. If it has bugs, you can edit, compile, and test fixes. I can liaise with the BOINC developers and pass back any successful fixes for other people to use.

You know, there are quite a few other people NOT having this problem. I'm not convinced there is a problem worth addressing.
As far as I know, None of my recommendations have been completed. I'm still waiting.

You know, there are quite a few other people NOT having this problem. I'm not convinced there is a problem worth addressing.
As far as I know, None of my recommendations have been completed. I'm still waiting.

I already do today. maybe you miss one of my post.

Dry the cache made some reschedules and see if anything happening
Nothing happening Back to work as normal
Was running the SSE4.1 Builds when i do the test.

But as i tried to explain, the issue is not happening all the time, i just notice 3 or 4 times in the last weeks.
All when the cache was running at the end.
And there is something different now, we don't have a mix of WU just blc.

Please not misunderstand i not know nothing about code just a user who try to test what you guyÅ› tell me to do.

Since i not believe it's the app itself I'm back to the AVX2 builds

If that happening again i will post.

---

OK, I was thinking it was about time to head out to the pub - I think we've about ground ourselves down to a halt for tonight, anyway.

I concur.

---

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

A proud member of the OFA (Old Farts Association)

The Major part of running the cache empty was to Delete the Client_state.xml once the cache was empty so a New Client_state .xml would be created. I suggest turning Off Ubuntu networking before deleting the file from the BOINC folder and removing the slot folders. With networking disabled, move both the Client_state files to the Root level of Home and delete the slot folders. Start BOINC so a new Client_state.xml will be created, then Stop BOINC and look at the new Client_state.xml and see if it has the correct <hostid></hostid> & <rpc_seqno></rpc_seqno>. Use the numbers from the old Client_state.xml for the new one, I think you know the drill. Then Run BOINC Normally with a reduced cache setting, Without any Rescheduling, and see if you have the same trouble.

If you do have the same trouble with BOINC 7.8.3, try it with BOINC 7.4.44 as mentioned previously.

Why on earth would anyome deliberately destroy all possible chance of recovering evidence? My comment to Jeff last night by PM was that the most likely outcome of following that advice was the dumping of all tasks, with them being marked 'abandoned' on the website.

I'm outa here - I've never liked the 'sledgehammer and two short planks' school of computer maintenance.

I guess you missed the part about running the cache empty...somehow.
Drink one for Me, cause I stopped drinking Years ago.

There's absolutely ZERO justification for trashing an existing client_state file based on the evidence accumulated here over the past two days, both from the actual incidents that have occurred in normal use and from the testing scenarios attempted so far. That's awful advice.

What is going to happen if the cache is EMPTY? Please do tell.
If NOTHING is going to happen, why is it Awful?
Why are you PMing about things that CAN'T Happen with an EMPTY CACHE.
Am I dealing with drunks here or what?
The Current Client_State file has been EDITED countless times, that is very good reason to try a New one, that hasn't been edited.

Please guy's not take that personal.

I i'm doing the kill of the client just to take out that from the questions.

Doing that now (i keep my cache very low).

Back ASAP

---

OK Done back with a new client_state.xml

Rebuilding the cache now.

Leave running AVX2. Any news i post.

---

OK Done back with a new client_state.xml

Just to verify that it truly is a "new" file, take a look in your current Event Log and make sure that there's no reference to a client_state_prev file. If it's there, it would appear shortly after startup.

I believe is, I stop the Boinc. Move both files files to the desktop. Check nothing was left. And restart the Boinc again.

I see no reference for the old file in the Event Log .

---

I believe is, I stop the Boinc. Move both files files to the desktop. Check nothing was left. And restart the Boinc again.

I see no reference for the old file in the Event Log .

Okay, since you moved both files out, BOINC wouldn't be able to find either. If you had just removed the current client_state file, BOINC would likely have just reverted to the client_state_prev file, which is essentially a backup.

Did you preserve hostid and rpc_seqno, as the ... as he said? Auto-lookup may have worked, but best of luck.

Edit - looks like you got away with it - preserved stats. Host 8396902. That could have been showing zero everywhere.

Did you preserve hostid and rpc_seqno, as the ... as he said? Auto-lookup may have worked, but best of luck.

Yes i know about that. I recover all the info of my host. Not want to lose my nice #2 Place. LOL

---

Thanks Juan. It will be helpful to know if you still have the same trouble with both BOINC 7.8.3 & the older 7.4.44.

Hope that could take some time.
To replicate the problem in real life i will going to wait for the next unscheduled outage or server problem to dry my caches and
Not see the problem appearing after only blc WU where crunching.
To you and all others that could sound idiot but i suspect the "shutdown timing problem" appears when only arecibo vlars are crunching on the GPU & CPU at least all the times when i noticed it was after that happening, at the end of my cache all the WU are Arecibo Vlars and i crunch them on the CPU & GPU since they crunch very fast on my GPU's too.
Will keep working using the rescheduled to try to pass the weekly outage and if the issue happening again will post & try to change to the older boinc.
Anyway i suggest to the devs to look why the GPU apps works with the lockfile and the CPU no. And if possible find a way to correct that.
The hosts & GPUs are each day become faster and maybe today We two only have the problem but who knows the future.
From my side i'm happy now. Now i know why the issue happens and how to bypass and keep my host running.

Thanks Jeff, Richard, TBar, Brent, Keith and all others who help me with this issue.

<edit> Now it's time for few sundays beers. Cheers.

---