I had 4 downloads stalled this morning, but no files remaining to be transferred as per the transfers tab. Aborted these WUs and reset the project.
Now 2 long-runs are incoming, but download of libcufft.so.8.0 is hanging again and again, transient HTTP error :-(

Well, libcufft.so.8.0 is a shocking 150 MB...

Well, libcufft.so.8.0 is a shocking 150 MB...

Oh my god, that's amazing ! It's too big for the internet ! I understand now why it's not working :D
But, if we use an african swallow ??!! (we can put an USB drive with the file into the coconut)

I have a 100down/10up Mbit FTTH link, downloading updates, OS images, etc at 5-10MB/s, no problem...

For the last 10 minutes libcufft downloads at just 30kbyte/s, feels quite 2002ish ;-)

For the last 10 minutes libcufft downloads at just 30kbyte/s, feels quite 2002ish ;-)

Remember the "bip bip" of the old internet modem ;-)
Even when we post a new comment in this post, the page is slow to refresh.
The new server have some really big networking issue or wait I/O. May be both and with applicatons issue too.

I am acually a senior Unix/Linux Admin sys. If you need help....

Still here a lot of outages:



02.12.2016 12:19:08 | GPUGRID | Temporarily failed download of cufft32_65.dll: transient HTTP error
02.12.2016 12:19:08 | GPUGRID | Backing off 00:03:02 on download of cufft32_65.dll
02.12.2016 12:19:08 | GPUGRID | Started download of tcl84.dll
02.12.2016 12:29:48 | | Project communication failed: attempting access to reference site
02.12.2016 12:29:48 | GPUGRID | Temporarily failed download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-pdb_file: transient HTTP error
02.12.2016 12:29:48 | GPUGRID | Backing off 00:02:32 on download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-pdb_file
02.12.2016 12:29:48 | GPUGRID | Started download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-psf_file
02.12.2016 12:29:50 | | Internet access OK - project servers may be temporarily down.
02.12.2016 12:34:55 | GPUGRID | Temporarily failed download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-psf_file: transient HTTP error
02.12.2016 12:34:55 | GPUGRID | Backing off 00:02:59 on download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-psf_file
02.12.2016 12:34:55 | GPUGRID | Started download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-par_file
02.12.2016 12:34:56 | | Project communication failed: attempting access to reference site
02.12.2016 12:34:57 | | Internet access OK - project servers may be temporarily down.
02.12.2016 12:34:57 | GPUGRID | Finished download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-par_file
02.12.2016 12:34:57 | GPUGRID | Started download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-conf_file_enc
02.12.2016 12:34:58 | GPUGRID | Finished download of e118s11_e24s4p0f4-
02.12.2016 12:44:39 | GPUGRID | Temporarily failed download of cufft32_65.dll: transient HTTP error
02.12.2016 12:44:39 | GPUGRID | Backing off 00:05:24 on download of cufft32_65.dll
02.12.2016 12:44:39 | GPUGRID | Started download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-pdb_file
02.12.2016 12:44:41 | | Project communication failed: attempting access to reference site
02.12.2016 12:44:41 | GPUGRID | Finished download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-pdb_file
02.12.2016 12:44:41 | GPUGRID | Started download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-psf_file
02.12.2016 12:44:42 | | Internet access OK - project servers may be temporarily down.
02.12.2016 12:44:42 | GPUGRID | Finished download of e118s11_e24s4p0f4-SDOERR_CASP22SnoS_crystal_ss_contacts_1ns_a3D_2-0-psf_file
02.12.2016 12:49:29 | GPUGRID | Started download of cufft32_65.dll
02.12.2016 12:54:41 | | Project communication failed: attempting access to reference site
02.12.2016 12:54:41 | GPUGRID | Temporarily failed download of cufft32_65.dll: transient HTTP error
02.12.2016 12:54:41 | GPUGRID | Backing off 00:08:00 on download of cufft32_65.dll
02.12.2016 12:54:43 | | Internet access OK - project servers may be temporarily down.
02.12.2016 12:55:07 | GPUGRID | Started download of cufft32_65.dll

New task, always 15ko/s for libtiff.
The good side: no problem for my torrents download.... My FTTH bandwich isn't worried... ;-) My RAC is !
File on a very old tape device ??

Edito: 1h47 => 66% I believe !

Does GPUGRID need a faster internet connection to make the new server effective?

New task, always 15ko/s for libtiff.
The good side: no problem for my torrents download.... My FTTH bandwich isn't worried... ;-) My RAC is !
File on a very old tape device ??

Edito: 1h47 => 66% I believe !

Download, finished and validated. More time to download than calcul, but, it is ok.

Does GPUGRID need a faster internet connection to make the new server effective?

+1
I'll second that question.

---

I would love to see HCF1 protein folding and interaction simulations to help my little boy... someday.

Does GPUGRID need a faster internet connection to make the new server effective?

+1
I'll second that question.

I'm not convinced that speed is the problem - or that GPUGrid's network speed is the problem.

I would observe that there is at least one, possibly two distinct, problem(s).

Maybe Spain's intranet has poor interconnectors with some other countries or some other ISPs? Some people clearly do experience slow connections, but I don't see that here in the UK - certainly not in the form of extended, steady, continuous trickles of data at low speed.

Instead, I see bursts of data: fast transmission, but with pauses. I mentioned a large file upload at ~850 KByte/sec last night: that's exceptionally good (almost the maximum rate my fibre ISP connection allows), but I usually get about half that rate, which is still pretty good.

I do observe the stalled downloads that we've discussed at length in other threads - and if not retried quickly enough, they can report as a start-to-finish average of rubbish. But the data flow, while it's flowing, is at a good instantaneous rate (most files are too small to register a meaningful average - it's been a long time since I needed to download one of the big cufft DLLs).

I was wondering if the new server would reveal that the download stalls had been a quirk of the old server: it hasn't, I've had the same stalled downloads today. So we need to look a little bit further out into the University or national infrastructure. The hunt restarts here.

My uploads have been steady and good here in the US. It's the downloads I'm having trouble with. As you said, the speed is good so long as it continues, but it doesn't. I have about 30 files to download with 8 partials waiting to start again. I had some yesterday about 8MB that took a few hours to download in about 25 small pieces when I kept clicking Retry Now. At times it was getting 300-400 KBps but only for a couple of seconds when it stalls.

Maybe someone can get on a speed test website and choose a server here in the US to see how uploads and downloads perform. I've used Speedtest.net that works well here. You're able to move the green box to see servers world wide. I'm just curious if it will work from there connected to servers here.

According to the website the server is in Barcelona. I got on Speedtest.net and chose a server there. My speeds were: download 12.86Mbps and upload 0.85MBps. Both went without hesitation or interruption. Ping was 198ms.

Any ideas? My graphics cards are just sitting here doing nothing for a couple of hours waiting for downloads that are pending....

I don't think that you can generalize. But just for fun I have tried the same from here (Austria) to Spain. 36,4MBit Download and 7,3MBit upload, 72ms Ping. So that is pretty different from the below and both results not very meaningful I assume. As if you want to "check" the bandwidth from Ireland to the Texas, whatever supposed to mean.

I think that the local Admin at the institute is the only one who can get the real numbers.

---

I would love to see HCF1 protein folding and interaction simulations to help my little boy... someday.

Just some further observations from today...

After setting <http_transfer_timeout>60</http_transfer_timeout>, I managed to download the libcufft, hitting retry every few minutes...
Since BOINC was showing an active transfer, but no increasing % or size of the file, I ran du -skx against it on disk every second:

Fr 2. Dez 10:25:11 CET 2016 108852 libcufft.so.8.0
Fr 2. Dez 10:25:12 CET 2016 109908 libcufft.so.8.0
Fr 2. Dez 10:25:13 CET 2016 112452 libcufft.so.8.0
Fr 2. Dez 10:25:14 CET 2016 116932 libcufft.so.8.0
Fr 2. Dez 10:25:15 CET 2016 124624 libcufft.so.8.0
Fr 2. Dez 10:25:16 CET 2016 130296 libcufft.so.8.0
Fr 2. Dez 10:25:17 CET 2016 130296 libcufft.so.8.0
Fr 2. Dez 10:25:18 CET 2016 130296 libcufft.so.8.0
Fr 2. Dez 10:25:19 CET 2016 130296 libcufft.so.8.0
Fr 2. Dez 10:25:20 CET 2016 130296 libcufft.so.8.0

Within 5 seconds the size grew from 108852 to 130296kB, then the download was hanging again. So sometimes the performance is there, but not stable...

Upload so far was always ok.

One of my PCs seems to have successfully downloaded, crunched, and returned a work unit post server upgrade.

I thought someone might be interested in the success...

---

My speedtest.net results to Barcelona:

hosted by | ping | down | up
          | msec |   Mbit/s  
Orange    |   67 |  40  | 10
CSUC      |   62 |  47  | 14
Eurona    |   56 |  51  | 15
Adamo     |   84 |  25  |  8

My bandwidth is 51.5/15.7 Mbit/s

Morning all
I can not receive any work tasks, since last Tuesday
any help would be good thanks

It is not an ISP problem: no problem before the "update".
I can imagine they use virtualization but with not the knowledge. Network virtualization can be really tricky I have to admit (am Admin Sys).

I proposed help but no reply. We have no communication from them in fact. Just "Toni" for https, Toni who is project administrator/dev/scientist. You make cofee too ?? (joke, no offense) Now it is the week-end so I expect nothing until monday...

I can not receive any work tasks, since last Tuesday

Nobody, a server update (1 day) and after problems (2 days).. They don't want to step back, so, waiting they (or her/he) solve them.

One of my PCs seems to have successfully downloaded, crunched, and returned a work unit post server upgrade.

I thought someone might be interested in the success...

You'ld think wouldn't you, but NO, :-)