Download size now 200 MB


Advanced search

Message boards : Number crunching : Download size now 200 MB

Author Message
computezrmle
Send message
Joined: 29 Oct 14
Posts: 54
Credit: 1,137,404
RAC: 0
    
Message 5880 - Posted: 26 Dec 2016, 11:39:34 UTC

Since 2016-12-23 the typical download size of an ATLAS WU changed from 50 MB to 200 MB.
Is this due to a misconfiguration of the current batch or do users have to accept this as normal now?

Erich
Send message
Joined: 18 Dec 15
Posts: 253
Credit: 1,942,248
RAC: 0
    
Message 5882 - Posted: 27 Dec 2016, 14:42:34 UTC

They seem to have reverted back to the previous size of tasks.
The tasks I downloaded today were only a few MBs.

computezrmle
Send message
Joined: 29 Oct 14
Posts: 54
Credit: 1,137,404
RAC: 0
    
Message 5883 - Posted: 27 Dec 2016, 15:19:51 UTC

Today (so far) I got 6 ATLAS jobs.
None of them less than 200 MB initial download.

[27/Dec/2016:03:06:45 +0100] "GET http://atlasathome.cern.ch/ATLAS/download/284/zAxNDm5wifpnDDn7oo6G73TpABFKDmABFKDmoOLKDmABFKDmkb1ESm_EVNT.10108100._000136.pool.root.1 HTTP/1.1" 200 217094456 "-" "BOINC client (x86_64-pc-linux-gnu 7.6.31)" TCP_MISS:HIER_DIRECT
[27/Dec/2016:04:30:37 +0100] "GET http://atlasathome.cern.ch/ATLAS/download/225/zcvMDmg4qfpn7jp7oou28CBqABFKDmABFKDmO5GKDmEqLKDm6a1uLo_EVNT.10108100._000150.pool.root.1 HTTP/1.1" 200 214624742 "-" "BOINC client (x86_64-pc-linux-gnu 7.6.31)" TCP_MISS:HIER_DIRECT
[27/Dec/2016:08:34:38 +0100] "GET http://atlasathome.cern.ch/ATLAS/download/1dd/RdzKDmkKufpnDDn7oo6G73TpABFKDmABFKDmevFKDmABFKDmUIt0Do_EVNT.10108100._000152.pool.root.1 HTTP/1.1" 200 213797029 "-" "BOINC client (x86_64-pc-linux-gnu 7.6.31)" TCP_MISS:HIER_DIRECT
[27/Dec/2016:08:35:12 +0100] "GET http://atlasathome.cern.ch/ATLAS/download/af/wd1MDmkKufpnDDn7oo6G73TpABFKDmABFKDmbwFKDmABFKDmUIt0Do_EVNT.10108100._000152.pool.root.1 HTTP/1.1" 200 213797028 "-" "BOINC client (x86_64-pc-linux-gnu 7.6.31)" TCP_MISS:HIER_DIRECT
[27/Dec/2016:09:05:31 +0100] "GET http://atlasathome.cern.ch/ATLAS/download/271/DEALDmV9ufpnDDn7oo6G73TpABFKDmABFKDmtcFKDmABFKDmmilh1n_EVNT.10108100._000154.pool.root.1 HTTP/1.1" 200 214944244 "-" "BOINC client (x86_64-pc-linux-gnu 7.6.31)" TCP_MISS:HIER_DIRECT
[27/Dec/2016:14:45:07 +0100] "GET http://atlasathome.cern.ch/ATLAS/download/2bb/lhrKDmJS1fpnDDn7oo6G73TpABFKDmABFKDmgRHKDmABFKDmNxDxwm_EVNT.10108100._000164.pool.root.1 HTTP/1.1" 200 217031083 "-" "BOINC client (x86_64-pc-linux-gnu 7.6.31)" TCP_MISS:HIER_DIRECT

Erich
Send message
Joined: 18 Dec 15
Posts: 253
Credit: 1,942,248
RAC: 0
    
Message 5884 - Posted: 27 Dec 2016, 15:58:47 UTC - in response to Message 5882.

... The tasks I downloaded today were only a few MBs.

Well, my statement came too early. The latest two tasks downloaded were again 205MB size ea.

For my cable modem no problem though at all.

rbpeake
Send message
Joined: 27 Jun 14
Posts: 86
Credit: 8,794,961
RAC: 0
    
Message 5894 - Posted: 29 Dec 2016, 18:41:49 UTC - in response to Message 5884.

... The tasks I downloaded today were only a few MBs.

Well, my statement came too early. The latest two tasks downloaded were again 205MB size ea.

For my cable modem no problem though at all.

This is not a problem for me, either, but maybe it makes sense for the project developers to make these work units run more iterations, so not so much relative time is spent on downloading?

computezrmle
Send message
Joined: 29 Oct 14
Posts: 54
Credit: 1,137,404
RAC: 0
    
Message 5897 - Posted: 29 Dec 2016, 22:03:41 UTC

One more reason to keep the initial download small.
In case of a rising error rate the server starts it´s own "DoS".
See: https://lhcathome.cern.ch/ATLAS/forum_thread.php?id=641&postid=5895

Profile MAGIC Quantum Mechanic
Avatar
Send message
Joined: 4 Jul 14
Posts: 332
Credit: 485,372
RAC: 0
    
Message 5898 - Posted: 30 Dec 2016, 5:32:04 UTC - in response to Message 5897.
Last modified: 30 Dec 2016, 5:32:52 UTC

I decided to load a few tasks here today on one of mine but when I checked the *transfers* I saw it wanted me to d/l OVER 1 GB ......well my dsl isp makes everything else a problem if I try doing that much as all the other Cern tasks I do every day.......so I had to give them the boot with a *reset*

1GB is too much
____________

computezrmle
Send message
Joined: 29 Oct 14
Posts: 54
Credit: 1,137,404
RAC: 0
    
Message 6015 - Posted: 18 Jan 2017, 13:35:55 UTC - in response to Message 5880.

Since 2016-12-23 the typical download size of an ATLAS WU changed from 50 MB to 200 MB.
Is this due to a misconfiguration of the current batch or do users have to accept this as normal now?

Initial download size is still more than 200 MB per WU.
Any comments from the project team?

Profile HerveUAE
Avatar
Send message
Joined: 18 Dec 16
Posts: 44
Credit: 509,829
RAC: 0
    
Message 6099 - Posted: 29 Jan 2017, 18:33:12 UTC - in response to Message 6015.

Initial download size is still more than 200 MB per WU.

I just downloaded 4 multi-core WUs, and each one was just a couple of MBs. I had not seen that since I joined mid Dec 2016.
____________
We are the product of random evolution.

computezrmle
Send message
Joined: 29 Oct 14
Posts: 54
Credit: 1,137,404
RAC: 0
    
Message 6102 - Posted: 30 Jan 2017, 18:14:23 UTC

During the last few days the progress bar for batch 10327233 had disappeared from the MainPage and the server obviously sent out WUs from other batches with much smaller initial data sets.
As of today the progress bar for 10327233 returned on the Main Page and so did the distribution of the 200 MB data sets.

I hope that future batches will produce much smaller initial data sets.

Message boards : Number crunching : Download size now 200 MB