LHC@home

LHCb result file was nearly 1GB!
:-O

[08/Apr/2018:19:30:17 +0200] "CONNECT lbboinc01.cern.ch:9148 HTTP/1.0" 200 994685415 "-" "-" TCP_TUNNEL:HIER_DIRECT

Found some more of that size in the logs:
1.3 GB
1.1 GB
1.0 GB
1.1 GB
1.2 GB

Will this become standard now?
What should volunteers with lower bandwidth do?

[07/Apr/2018:13:48:37 +0200] "CONNECT lbboinc01.cern.ch:9148 HTTP/1.0" 200 1267818530 "-" "-" TCP_TUNNEL:HIER_DIRECT
[07/Apr/2018:17:02:46 +0200] "CONNECT lbboinc01.cern.ch:9148 HTTP/1.0" 200 1085743755 "-" "-" TCP_TUNNEL:HIER_DIRECT
[09/Apr/2018:01:29:42 +0200] "CONNECT lbboinc01.cern.ch:9148 HTTP/1.0" 200 997377919 "-" "-" TCP_TUNNEL:HIER_DIRECT
[11/Apr/2018:09:36:35 +0200] "CONNECT lbboinc01.cern.ch:9148 HTTP/1.0" 200 1140769680 "-" "-" TCP_TUNNEL:HIER_DIRECT
[11/Apr/2018:13:35:39 +0200] "CONNECT lbboinc01.cern.ch:9148 HTTP/1.0" 200 1188767940 "-" "-" TCP_TUNNEL:HIER_DIRECT

Found some more of that size in the logs:
1.3 GB
1.1 GB
1.0 GB
1.1 GB
1.2 GB

that's interesting - I havn't had any such huge one yet

BTW, it's about a factor of 10 compared to the normal result size.

It's climbing to new heights and happens on different hosts:
1.2 GB
1.3 GB
1.4 GB
1.6 GB

Is it due to a configuration error or did some tasks "escape" that were thought to be processed inside the CERN network?
Comments from the project team would be appreciated.

[12/Apr/2018:06:24:16 +0200] "CONNECT lbboinc01.cern.ch:9148 HTTP/1.0" 200 1197658388 "-" "-" TCP_TUNNEL:HIER_DIRECT
[13/Apr/2018:10:45:15 +0200] "CONNECT lbboinc01.cern.ch:9148 HTTP/1.0" 200 1255994250 "-" "-" TCP_TUNNEL:HIER_DIRECT
[13/Apr/2018:11:59:37 +0200] "CONNECT lbboinc01.cern.ch:9148 HTTP/1.0" 200 1447436119 "-" "-" TCP_TUNNEL:HIER_DIRECT
[13/Apr/2018:22:21:42 +0200] "CONNECT lbboinc01.cern.ch:9148 HTTP/1.0" 200 1550429788 "-" "-" TCP_TUNNEL:HIER_DIRECT

During the past 2.5 weeks there were 78 uploads with more than 0.5 GB each.
That sums up to nearly 80 GB.

What should volunteers with lower bandwidth do?

Guess.

LHCb big result statistics between 2018-05-01 and 2018-08-11


Month
total_results/big_results_(>200MB)/%
max_upload_size


May
3122/18/0.6%
1.86 GiB

June
2739/36/1.3%
1.29 GiB

July
3889/45/1.2%
0.96 GiB

August
1424/88/6.2%
1.01 GiB


Please avoid sending out LHCb tasks to private volunteers that produce result files of that size.

Statistics as of last weekend 2018-08-11/12.

total_results/big_results_(>200MB)/%
max_upload_size

267/64/24.0%
1.01 GiB

Have you seen the same in -dev. Because there is a Multithread-Program?

Had only 3 valid LHCb WUs at -dev (May 2018).
It would be lots of work to extract the relevant information from the logfiles based on the timestamps.

As there were no changes regarding my software environment it's most likely an issue at CERN.

Have two-Core running in -dev (2 Tasks at the same moment) since weeks without problems!

There is no real problem nor any error message except the upload size of the (intermediate) result files which is occasionally 10 times as much as usual.
All of my last >120 LHCb WUs from different hosts finished with success.

I also have them running ok here and the 2-core version over at -dev
(Atlas is a waste of time for me since they take 10 hours to d/l and 4hrs at most to run)