RAL Tier A Status - Science and Technology Facilities Council

Download Report

Transcript RAL Tier A Status - Science and Technology Facilities Council

RAL Tier A Status
Tim Adye
Rutherford Appleton Laboratory
BaBar UK Collaboration Meeting
Royal Holloway
19th September 2003
19th September 2003
Tim Adye
1
136 M
20 May
27 Ma y
Ma y
103 J a y
17 Juun
24 Ju n
1Ju nn
J
158 Jul
22 Juul
29 Ju l
125 AJu ll
19 Aug
26 Aug
2 Auug
S g
169 Sep
23 Sep
30 Sep
Sep
147 Oep
21 Oct
28 Oct
Oc
114 N ctt
18 No v
25 Nov
2 Noov
D
169 De cv
23 De c
30 Dec
Dec
136 Jec
20 Jaan
27 Ja n
n
103 FJa n
17 Feb
24 Feb
Feb
103 Meb
17 Mar
24 Ma r
31 Ma r
Ma r
147 Aa r
21 Apr
28 Apr
p
125 MAprr
19 May
26 Ma y
a
2 Ma y
J y
169 Jun
23 Juun
30 Ju n
Ju n
147 Jun
21 Jul
28 Ju l
114 AJu ll
18 Aug
25 Aug
u
1 Aug
8 Seg
Sep
p
(Normalised to P450)
BaBar CPU Hours per Week
BaBar Batch CPU Use at RAL
Full usage at full efficiency of BaBar CPUs = 106,624 Hours/Week; 59,733 according to MOU
140,000
120,000
100,000
80,000
60,000
40,000
20,000
0
Week Beginning
19th September 2003
Tim Adye
SP
UK Users
Non-UK Users
2
Farm Usage
19th September 2003
Tim Adye
3
163 M
20 M ay
27 May
May
103 Juay
17 J n
24 Juun
n
1Jun
J
158 Jul
22 Juul
29 Jul
125 AJul
19 A ugl
26 A ug
2 Auug
S g
196 Sep
23 Sep
30 S ep
S ep
147 Oep
21 O ct
28 Oct
c
114 NOctt
18 N ov
25 N ov
2 Noov
D
196 Decv
23 Dec
30 D ec
D ec
163 Jec
20 Jaan
27 Jan
n
103 FJan
17 F eb
24 F eb
F eb
103 Meb
17 M ar
24 Mar
31 Mar
Ma
174 A arr
21 Apr
28 A pr
p
152 MAprr
19 M ay
26 May
a
2May
Juy
9
16 J n
23 Juun
30 Jun
J n
147 Juun
21 Ju l
28 Jul
114 AJul
18 A ugl
25 A ug
1 Auug
8 Se g
S ep
p
BaBar Users per Week
BaBar Batch Users at RAL
50
A total of 229 new BaBar users registered since December 2001
(running at least one non-trivial job each week)
45
40
35
30
25
20
15
10
5
0
Week Beginning
19th September 2003
Tim Adye
UK Users
Non-UK Users
4
Recent Changes
• New BaBar CSF system manager, Martin Bly
• replaces Phil Radden
• New RAL BaBar (+Grid) RA, Chris Brew
• New CSF hardware arrived in March
• 80 dual Pentium 4 2.66 GHz (2GB RAM)
• Currently allocated to other users, leaving the entire “old”
farm just to us (286 P1400, 22 P1000)
• 11 disk servers, each with 2 x 1.8 TB file-systems
• 38 TB total now allocated to BaBar – all full!
19th September 2003
Tim Adye
5
CPU Usage for all CSF groups
19th September 2003
Tim Adye
6
Disk allocation for all CSF groups
19th September 2003
Tim Adye
7
Recent Changes (cont)
• Batch worker nodes upgraded to RedHat 7.3
• Will simplify merging BaBar tier1a with UK tier1 farms
• should allow us to “borrow” spare capacity
• Front-ends remain RedHat 7.2
• analysis-13b (10.4.4) and analysis-14(a) (12.5.2)
compiled in RH72, but jobs validated to run on RH73 as
well
• Last RedHat 6 box switched off on Tuesday
• “Safer” Kanga conditions updating procedure (Chris)
• Linux Objectivity server installed and running
• SP
• Objy Conditions for series-12
19th September 2003
Tim Adye
(Wahid)
(TimB)
8
Other Planned Changes
• Finally bring old Sun disk server back into service
• Reconfigured and extensively tested since catastrophic
file-system problems in February
• Will provide 4.2 TB of sorely needed space
• Try to keep older data there to minimise risk/load
• CVS hack to allow checkouts (addpkg) without AFS
token
• May require a minor change at other sites that use RAL
CVS – Manny will post details soon
• Move user staging disk (/stage/babar-user1 aka
$BFROOT/work) to larger and faster disk
(Manny)
• 0.8 TB  1.4 TB
19th September 2003
Tim Adye
9
Kanga Data at RAL
• All available disk is now full
• 4 TB more on the Sun in next day or two, but that will be
quickly filled
• Need space for run 3 and series-12 reprocessing
• SP signal and generic conversion going well
• Real data conversion has to be restarted
• AllEvents and 15 streams for data and generic MC
• Need space for New Kanga (CM2) staging area
• Most will be kept in the Datastore (tape) and staged in
automatically
• Need to remove unused data
• “Bad” series-12 real data
• Series-10 old tag (<=K10.4.1b)
• Anything else?
19th September 2003
Tim Adye
10
Support
For help, post to “RAL Tier A” HyperNews forum
or contact
• Emmanuel Olaiya (at SLAC) or
• Chris Brew or me (at RAL)
19th September 2003
Tim Adye
11