Your university or experiment logo here Tier1 Deployment Steve Traylen.
-
Upload
matthew-thomas -
Category
Documents
-
view
212 -
download
0
Transcript of Your university or experiment logo here Tier1 Deployment Steve Traylen.
![Page 1: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/1.jpg)
Your university or experiment logo here
Tier1 Deployment
Steve Traylen.
![Page 2: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/2.jpg)
Your university or experiment logo here
Contents
• Job Efficiencies• CPU Fairshare• Installation Method at the Tier1.• Recent/Upcoming Services.
![Page 3: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/3.jpg)
Your university or experiment logo here
Utilisation
CPU Use (KSI2K*CPUMonths)
0
100
200
300
400
500
600
700
800
900
J an Feb Mar Apr May J un J ul Aug Sep Oct Nov Dec
Other
SNO
Zeus
H1
Minos
UKQCD
LHCB
DZERO
CMS
CDF
Babar
Atlas
Alice
![Page 4: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/4.jpg)
Your university or experiment logo here
Efficiencies.
• New plots are now generated regularly.– http://www.gridpp.rl.ac.uk/stats/
• A report has been written for the Tier1 Board that describes how these are being calculated.– http://www.gridpp.ac.uk/tier1a/board/
doc/Utilisation_OC_August_2005_V1.0.doc
![Page 5: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/5.jpg)
Your university or experiment logo here
Overall, (total cpu/total wall)
![Page 6: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/6.jpg)
Your university or experiment logo here
Individual VOs
![Page 7: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/7.jpg)
Your university or experiment logo here
Individual Jobs
![Page 8: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/8.jpg)
Your university or experiment logo here
Fair Share Policies
• Currently it is completely based on UNIX groupid.
• Any spare capacity is allocated to over subscribers in proportion to their allocation.
• This may change…..
![Page 9: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/9.jpg)
Your university or experiment logo here
Fair Share Policies (2)
• We can create accounts in Maui.– e.g. an LHC account and a non-LHC account.– e.g. a grid account and a non Grid account.
• These accounts can be given allocations.– But we now have competing scheduling
regimes.– Each of the regimes can be weighted but the
initial values and subsequent tuning is not obvious.
![Page 10: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/10.jpg)
Your university or experiment logo here
Fair Share(3)
• New plots are now being constructed.• They will plot:
– allocated % for different VOs.– utilised % for different VOs.– allocated/utilised for different VOs.
![Page 11: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/11.jpg)
Your university or experiment logo here
How is tier1 deployed.
• Still done with.– DHCP/PXE/Kickstart and bunch of scripts.
• YAIM is just one bit.
– Local config packages are built.• e.g. tier1-ntp-config, tier1-sendmail-config, tier1-
lmsensors-config, tier1-yaim-config,….
– Packages maintained with yum/yumit.
• Batch workers:– Maintained via parallel SSH.– Down nodes are reinstalled routinely.
• Complicated service nodes:– Maintained by hand, often the only way.– Backed up to tape.
![Page 12: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/12.jpg)
Your university or experiment logo here
How the tier1 is deployed.
• It is simple especially for new starters.– Only bash and rpm building is required.– Flexible, extending what is there is easy.
• Problems– Things can get out step, though in reality
not a huge problem.• Configuration in packages helps a lot.
– We changed our netmask everywhere via a package upgrade.
• Farm wide configuration changes are rare.
![Page 13: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/13.jpg)
Your university or experiment logo here
Yumit/Pakiti
• http://sf.net/pakiti/• Yumit renamed as Pakiti
– Now support yum/up2date/apt-get.
• Improvements are being added now.– Currently only packages that could be
updated published.– It will also display installed packages.
• This will flag a WN with a missing/extra RPM.• Not a security feature but very useful.
![Page 14: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/14.jpg)
Your university or experiment logo here
Recent Deployments.
• FTS v1.3 is now being deployed.– Install complete by Matt Hodges– Understanding is a lot more thorough than
with FTS v1.1.2 when it was thrown up.
• VOBox primarily for Alice.– Install complete by Catalin Condurache.– Testing/Validation with EIS group
underway.– Bugs found – new version released today.
![Page 15: Your university or experiment logo here Tier1 Deployment Steve Traylen.](https://reader035.fdocuments.in/reader035/viewer/2022070412/5697bf751a28abf838c8063b/html5/thumbnails/15.jpg)
Your university or experiment logo here
Upcoming Deployments
• The two most critical boxes at RAL.– PBS scheduler.
• Recent crashes have stopped submissions for a day or two at least twice in last three months.
– DCache head node.• Work is now starting on use of the HA-
Linux software package to increase redundancy.– http://www.linux-ha.org/– Other (big) sites in the UK will probably want
this.