slow downloads?


log in

Advanced search

Message boards : Number crunching : slow downloads?

Author Message
Profile alephnull
Volunteer tester
Send message
Joined: 16 Mar 03
Posts: 119
Credit: 161,955,702
RAC: 369
United States
Message 1250498 - Posted: 23 Jun 2012, 17:36:54 UTC

sorry for the bother but i was wondering if others may be having issues with slow downloads recently? when i was running boinc version 6.10.58, the issues were not as bad as they seem now. after experimenting with versions 6.10.58, 6.12.34, 7.0.25 and currently 7.0.28 (test version), i was finally able to at least get some work for seti on this host. it took a little playing around because that particular host has quite a lot of einstein work so the scheduler would not send anything for seti.

currently for that host, i have 118 wu downloading and 4 ready to start (for seti).

for the work units downloading, the download will start, hover around 0.4 - 2 Kbs, then backoff. in the past 30 minutes or so ive probably been able to complete about 10 or so seti wu downloads.

just wondering if anyone else experiencing really slow downloads currently? sorry to ask a question thats been asked millions of times. just seems bad in the past couple of days. i would like to have the machine mentioned above a dedicated gpu seti cruncher but i cant get the work it seems, let alone build a sufficient work buffer (hence the einstein work).

currently running boinc 7.0.28. any info/comments are appreciated.
____________

Claggy
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 4048
Credit: 32,693,420
RAC: 508
United Kingdom
Message 1250505 - Posted: 23 Jun 2012, 17:43:07 UTC - in response to Message 1250498.
Last modified: 23 Jun 2012, 17:48:40 UTC

The link is maxed out at the moment, as the Cricket Graph shows:

Graphs for gigabitethernet2_3

At the moment there are two download servers, DNS switches between them every 5 minutes, by suspending, then resuming your network activity, may get you a faster download.

Claggy

Profile alephnull
Volunteer tester
Send message
Joined: 16 Mar 03
Posts: 119
Credit: 161,955,702
RAC: 369
United States
Message 1250512 - Posted: 23 Jun 2012, 17:46:53 UTC - in response to Message 1250505.

ok, ill just wait a couple of days and see if the work buffer fills up eventually. thanks for the info.
____________

Cosmic_Ocean
Avatar
Send message
Joined: 23 Dec 00
Posts: 2238
Credit: 8,453,224
RAC: 4,095
United States
Message 1250525 - Posted: 23 Jun 2012, 18:30:06 UTC

Wait.. I know there were two download servers.. but didn't it get consolidated down to one recently?

At any rate, since the pipe to the world is usually maxed out, downloads will be slow and may time-out and get dropped/lost packets on occasion. It will work itself out in the end, but you just have to be patient and wait.

For instance, when there are tapes of raw data available for the AstroPulse splitters to chew on, it takes me 3-10 days to build a 10-day cache, which is only ~65 WUs. Just have to wait and it will eventually happen.
____________

Linux laptop uptime: 1484d 22h 42m
Ended due to UPS failure, found 14 hours after the fact

Claggy
Volunteer tester
Send message
Joined: 5 Jul 99
Posts: 4048
Credit: 32,693,420
RAC: 508
United Kingdom
Message 1250534 - Posted: 23 Jun 2012, 18:50:46 UTC - in response to Message 1250525.

Wait.. I know there were two download servers.. but didn't it get consolidated down to one recently?

According to Matt's lastest Tech News georgem is hosting the Workunit files:

In better news, georgem is now not only hosting all the workunits and running some backend BOINC services and scientific analysis processes, but it's also hosting all the data (~13TB) from a recent survey of the galactic center collected at Green Bank Telescope. Several grad students will be processing this data on georgem itself.

and Slavac's status update says some more, but there's still some tuning to do before Vader and Anakin are retired:

Georgem, attached to 24 drives in the 45 JBOD, is currently taking care of the following:

1. being the workunit storage server for the public project, as well as the workunit/result storage server for the beta project (thus replacing our old file server, gowron)
2. since the workunits are local, some BOINC backend processes are executed on georgem
3. various scientific processes are also running on georgem (thus replacing an old, flaky computer server, maul)
4. with the extra storage we keep various science database backups here as well (in a roundabout way replacing another old, flaky file server, worf)
5. we also hope to tune apache well enough to make georgem the eventual download server itself (thus replacing both vader and anakin)
6. and if #5 goes well we also hope to move the results for the public project onto georgem and make it also the upload server (thus replacing bruno)

In short, georgem is handling a lot at the moment, already let us retire a couple servers, and soon a couple more.


Claggy

Message boards : Number crunching : slow downloads?

Copyright © 2014 University of California