Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 

Quick Go »
No member browsing this thread
Thread Status: Active
Total posts in this thread: 21
Posts: 21   Pages: 3   [ Previous Page | 1 2 3 | Next Page ]
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 1936 times and has 20 replies Next Thread
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: No more immediate upload/download?

Can't ask you to edit the OP and insert [RESOLVED] since it's not your thread. Anyway, if you please edit into your previous post what the final version is you're up and running again [6.10.58 I'd guess], and lossless I understand, all work recovered, than next visitors know what to get :D

Happy crunching, again.

--//--
[Nov 25, 2011 4:02:52 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Ingleside
Veteran Cruncher
Norway
Joined: Nov 19, 2005
Post Count: 974
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: No more immediate upload/download?

Don't think anything is lost, yet, and good you confirm it was a proxy acting up.

Not sure how you managed to run a repair of 6.10.58 over 6.12.34. Normally if downgrading to a lower version it will run as as a normal install and replace the newer

BOINC has newer supported "upgrading" to an older client by just running the older clients install, this is a sure method to need to re-run the old installer afterwards and choose "repair".

To downgrade, it's always been:
1: Uninstall current version.
2: Install older version.

This method will in most instances work, as long as you're not switching between 32 and 64-bit clients that is. As already mentioned you can't downgrade from v6.13.5 and later clients to a v6.13.1 or earlier client.
----------------------------------------


"I make so many mistakes. But then just think of all the mistakes I don't make, although I might."
[Nov 25, 2011 5:00:59 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: No more immediate upload/download?

Can't remember having walked into the ''repair'' stumble on downgrading. Probably, when trying latest alpha / beta, took the prudent route and uninstalled first before putting the stable version back. So far this test-platform has been running with the same device ID at WCG since over 5 years, touch wood. Today I've installed 6.13.12 alpha over 6.12 and it went com d'abitude. Re-running the installer to get the *repair* screen of this version suggests there's a *modify* option in addition to repair and remove... but there is not. Has there ever been?

--//--
[Nov 25, 2011 7:24:42 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: No more immediate upload/download?

sorry for the delay. Thanksgiving holiday.

I am on the 6.10.58 client.
I run the app whenever the pc is on but I don't leave the computer on when I am asleep or at work. I currently have things set with 0.50 days of work cached.

It seems to work most of the time with 0.50 days of cache. The only negative is it means the computer is not always running 2 Clean Energy units. Instead it works 1 or 2 every 2-4 days since the system caches us the Cancer units and only pulls in the Clean Energy ones if 1 or 2 have already been done.

I was just wondering. I never had a problem with any of the older clients and could not find a topic anywhere on here so I figured I would see if anyone had experienced the issue. Maybe after some time if will calm back down.
[Nov 26, 2011 10:02:55 PM]   Link   Report threatening or abusive post: please login first  Go to top 
KerSamson
Master Cruncher
Switzerland
Joined: Jan 29, 2007
Post Count: 1680
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: No more immediate upload/download?

Since around 3 or 4 weeks, I noticed that hosts do not report anymore every 4 hours but mostly only once daily.
All hosts are configured with profile preferences (no local definition).
Although I performed several configuration reviews, I don't know what the reason for this behavior could be.
With 4 and 6 cores hosts, I experience regularly a reporting queue that could be 5 CPU days long.
It is not a real issue since all these WUs will be validated but it makes the validation process longer.
Yves
---
PS: I did not update the used boinc clients since one year.
----------------------------------------
[Nov 27, 2011 10:02:25 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: No more immediate upload/download?

Don't know what's changed [nothing, since your client sets the rules], but my 6.12.33 client still fetches work whenever there's room in the cache and then combines reports with that.

Network is permanently available, so start by setting the Activity menu to "Network always available" if not set to that already, and the "connect every ..." to 0.00 days. With larger caches, this undulation of TTC shrinking and expanding can cause clients to go from small to hugely inflated caches, which then when there are no work requests needed, postpones the reporting up to 24 hours (there are about 10 conditions that the client checks]. The larger the cache and the more variable the project task duration, the more chance such is observed. Just let it run and btw, there's no ''every 4 hours" rule, but some back-off counter. And, RtR's *ARE* cleared no later than 24 hours after completion (Did you mean 5 CPU days or 5 core days?]

The BOINC 7.0 client [far out next year release] has an interesting change that was discussed in development and is actually working to the favor of CEP2 crunchers with the alpha test client. It will postpone work requests during upload, then when the Ready to Report is cleared, a work fetch is combined which immediately pulls a new CEP2 job and fills remaining space with other work i.e. like Beta, if CEP2 is opted in, they are the first hoppers checked for supply [the repairs are in that same *first call* group too]. Then other science tasks are requested. It's a substantial change from 6.12 and before where RtR's are combined with a work fetch request. Certainly I'm seeing a greater CEP2 production on my duo that is set to only allow 1 at any one time [small cache, so they're started quickly].

The server side, an ability is added to flag an individual task / project for immediate reporting. The new client would listen to that instruction [don't think the old could]. Per knreed we're maybe 3 months away from that upgrade [taking that as optimistic and silently hoping that this feature is put into action permanently on CEP2]

--//--

** Testing the next client: Please do, but it's not backward compatible with 6.12 and earlier. Downgrading might loose you your client install ID. For support and problems: To be reported at the alpha mailing list.
[Nov 27, 2011 11:11:36 AM]   Link   Report threatening or abusive post: please login first  Go to top 
KerSamson
Master Cruncher
Switzerland
Joined: Jan 29, 2007
Post Count: 1680
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: No more immediate upload/download?

Did you mean 5 CPU days or 5 core days?

I mean CPU day = core day
Otherwise I would use "calendar day".
Likewise, CPU time vs. wall clock time.
Yves
----------------------------------------
[Nov 27, 2011 12:40:18 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: No more immediate upload/download?

In short, your RtR's for the 6 core device is less than 24 hours per core. That's fine, long as there are no work fetches. If there are successful work-fetches and no reporting, then there's something wrong, but until 2 minutes ago my last task *self* reported, all seems to sail fine (but maybe currently the profile change messenger.)

--//--
[Nov 27, 2011 12:49:11 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Ingleside
Veteran Cruncher
Norway
Joined: Nov 19, 2005
Post Count: 974
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: No more immediate upload/download?

The BOINC 7.0 client [far out next year release] has an interesting change that was discussed in development and is actually working to the favor of CEP2 crunchers with the alpha test client. It will postpone work requests during upload,

But only for 5 minutes after an upload has started (if it's working as it should), so atleast in my case a work-request can happen long before the CEP2 is finished uploaded.

then when the Ready to Report is cleared, a work fetch is combined which immediately pulls a new CEP2 job and fills remaining space with other work i.e. like Beta, if CEP2 is opted in, they are the first hoppers checked for supply [the repairs are in that same *first call* group too]. Then other science tasks are requested. It's a substantial change from 6.12 and before where RtR's are combined with a work fetch request. Certainly I'm seeing a greater CEP2 production on my duo that is set to only allow 1 at any one time [small cache, so they're started quickly].

The biggest change is that it's using "Connected..." as the low-cache-size and "Connected..." + "Additional..." as the high-cache-size, and only asks for work each time drops below the low-cache-size. Since WCG max sends 15 tasks in a single scheduler-reply (except for lost tasks that is), depending on users "Additional..."-setting the cache is often never full of work any longer, and in practice you'll likely to have around 15 tasks to report.

Example, on a quad-core there each task takes 6 hours to run, any "Additional..." > 1 day won't be fulfilled by WCG since the max 15 tasks is less than 1 days cache.

Oh, and for anyone using a zero "Connected...", the client won't ask for more work except if out of work on a core.

The server side, an ability is added to flag an individual task / project for immediate reporting. The new client would listen to that instruction [don't think the old could]. Per knreed we're maybe 3 months away from that upgrade [taking that as optimistic and silently hoping that this feature is put into action permanently on CEP2]

Per-task immediately reporting is included in v6.12.27 and later clients.
----------------------------------------


"I make so many mistakes. But then just think of all the mistakes I don't make, although I might."
[Nov 27, 2011 12:51:09 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: No more immediate upload/download?

Well, my duo over WIFI takes 7.3 minutes to upload the _4 task part. The behavior is consistent. Same time a CEP2 is reported a new one is fetched instantly. Amazingly, the downloading line appears even before the ack is registered in the message log, so I'm readin:

533 WCG 27-11-2011 08:03 Started upload of E204132_534_A.30.C23H13N3S4.705.4.set1d06_0_0
534 WCG 27-11-2011 08:03 Started upload of E204132_534_A.30.C23H13N3S4.705.4.set1d06_0_1
535 WCG 27-11-2011 08:03 Finished upload of E204132_534_A.30.C23H13N3S4.705.4.set1d06_0_0
536 WCG 27-11-2011 08:03 Started upload of E204132_534_A.30.C23H13N3S4.705.4.set1d06_0_2
537 WCG 27-11-2011 08:03 Finished upload of E204132_534_A.30.C23H13N3S4.705.4.set1d06_0_1
538 WCG 27-11-2011 08:03 Started upload of E204132_534_A.30.C23H13N3S4.705.4.set1d06_0_3
539 WCG 27-11-2011 08:03 Finished upload of E204132_534_A.30.C23H13N3S4.705.4.set1d06_0_2
540 WCG 27-11-2011 08:03 Finished upload of E204132_534_A.30.C23H13N3S4.705.4.set1d06_0_3
541 WCG 27-11-2011 08:03 Started upload of E204132_534_A.30.C23H13N3S4.705.4.set1d06_0_4
542 WCG 27-11-2011 08:08 [checkpoint] result GFAM_x1j3k_w2WATsNDP_0000423_0227_0 checkpointed
543 WCG 27-11-2011 08:10 Finished upload of E204132_534_A.30.C23H13N3S4.705.4.set1d06_0_4
544 WCG 27-11-2011 08:10 [sched_op] Starting scheduler request
545 WCG 27-11-2011 08:10 Sending scheduler request: To report completed tasks.
546 WCG 27-11-2011 08:10 Reporting 1 completed tasks, requesting new tasks for CPU
547 WCG 27-11-2011 08:10 [sched_op] CPU work request: 9804.51 seconds; 0.00 CPUs
548 WCG 27-11-2011 08:10 Scheduler request completed: got 1 new tasks
549 WCG 27-11-2011 08:10 [sched_op] Server version 601
550 WCG 27-11-2011 08:10 Project requested delay of 11 seconds
551 WCG 27-11-2011 08:10 [sched_op] estimated total CPU task duration: 46849 seconds
552 WCG 27-11-2011 08:10 [sched_op] handle_scheduler_reply(): got ack for task E204132_534_A.30.C23H13N3S4.705.4.set1d06_0
553 WCG 27-11-2011 08:10 [sched_op] Deferring communication for 11 sec
554 WCG 27-11-2011 08:10 [sched_op] Reason: requested by project
555 WCG 27-11-2011 08:10 Started download of 3fd423e341334320a4372ef96be2825e.zip
556 WCG 27-11-2011 08:10 Finished download of 3fd423e341334320a4372ef96be2825e.zip

"Connect every..." set to zero postponing work-fetching until cache is zero / core goes idle? Well so happens it's set to 0.01 with a 1.0 additional buffer and now running 2 days with "Run always", network set to "Always available", so you tell us how that tallies with the cache never dropping below the setting, but briefly... maybe it's because it's a duo never needing to cache 15 of the selected projects (there are 6 now making up about 28 hours per core), opposed to a sextuplet ;P. Somehow I'm seeing more update button hammering coming to the grid, more than is desired. So far I've not had to.

--//--
[Nov 27, 2011 1:24:37 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Posts: 21   Pages: 3   [ Previous Page | 1 2 3 | Next Page ]
[ Jump to Last Post ]
Post new Thread