BEIJING-LCG2 Site Report Dec. 2013 Beijing
2010.12-2011.11 2011.12-2012.11 2012.11-2013.10
Job Number
4,786,574 5,555,044 2,354,501
Walltime 4,869,401 6,783,696 4,079,095
CPUEfficiency
82.1% 80.8% 81.9%
Node Utility
74.9% 79% 48.2%
China User CPU Time
36,112 95,421 82,813
Resource:• Disk array retire. Add a new disk array. Totally 80TB
01/12 02/12 03/12 04/12 05/12 06/12 07/12 08/12 09/12 10/12
可用性 0.9865 0.9685 1 1 1 0.94 0.9408 0.9924 0.7565 0.9681
可靠性 1 0.97 1 1 1 1 0.9923 0.99 0.8617 0.96
5%
15%
25%
35%
45%
55%
65%
75%
85%
95%
2013Reliability And Availability BEIJING-LCG2
AnnualMaintain
s
ATLAS Available Disk
Grid Jobs statistics 2013
Jan Feb Mar Apr May Jun Jul Aug Sep Oct0
200,000
400,000
600,000
800,000
2013 Grid Jobs Statistics
2012 2013
CP
U T
ime (
Hour)
63.47%
34.51%
2.02%
2013 CPU Time
atlas cms Others
atlas biomed cms
80.1%72.2%
81.7%
91.2%82.2%
65.4%
Cpu Efficiency
2012 2013
EMI Upgrade July 2013 All Service upgrade to EMI2
Cream, dCache, DPM,APEL, BDII, WMS, LB Sep. 2013 SL6 Worknode Migragion Otc. 2013 dCache,DPM Upgrade to SHA2
compatible version Nov. 2013 Upgrade All host certificates. Nov. 2013 WMS upgrade EMI3 and SHA2
compatible version All service are SHA2 compatible version
System management tools upgrade Upgrade from quattor to Puppet+Foreman Foreman as external node configure of
Puppet Foreman as Puppet dashboard Puppet Modules need to develop:
Ntp、 pkg、 lustre、 afs-authlogin、 autofs、 account s、 symlinks、 sys-directorys、 pbs-clients、 cronjobs、 gluster、 sudoer、 sshconfig 、 ganglia-monitor、 profile_env、 nagios-nrpe、 security-limits
Near 100 hosts managed by Puppet+Forman.
Tire2 Site issue
Typical site incidents: Very old hard ware. All the disk array were 5 years
old. Disk array retire Middleware upgrade frequently. Need lot of work
to do on it. EMI2, EMI3, SHA2
ありがとうMerci
Thanks