110 likes | 230 Views
LAL Site Report. Michel Jouvin LAL / IN2P3 jouvin@lal.in2p3.fr. 100 Go. FC. FC. HDS 9570 4 TB. ESA12000 1,5 TB. Electronic CAD (Sun + Cadence). Alpha Experiments (8 CPUs). Linux Experiments (30 CPUs). GRID Fram (10 CPUs). Mac (100). Xterm. PC (300). Main Resources. DS20.
E N D
LAL Site Report Michel Jouvin LAL / IN2P3 jouvin@lal.in2p3.fr
100 Go FC FC HDS 9570 4 TB ESA12000 1,5 TB Electronic CAD (Sun + Cadence) Alpha Experiments (8 CPUs) Linux Experiments (30 CPUs) GRID Fram (10 CPUs) Mac (100) Xterm PC (300) Main Resources DS20 LSF Cluster • NFS • SMB • Appletalk • www • Mail • Print DS20e Gb Ethernet 100 Mb 10 Mb LAL Site Report - HEPix - Edinburgh 2004
Main Resources Changes • More Linux CPUs • 25 dual Opteron 2,2 on order (IBM E325, 1U) • More TBs • HDS 9570V : 2,5 TB added • Running very well • Less budget… • Budget uncertainty (always changing) • No OS upgrade since last meeting LAL Site Report - HEPix - Edinburgh 2004
Mail Service • Virus/Spam filtering production • Spam Assassin + MimeDefang • Tunning in progress to increase efficiency (~70%) • Late with all our main projects for mail service • SIEVE for filtering at message delivery (server based) • Upgrade IMAP server to Cyrus v2 • Required for SIEVE filtering • Authenticated SMTP • Conversion of our current SASL db for transparent user migration with same password db for IMAP and Sendmail LAL Site Report - HEPix - Edinburgh 2004
Windows Infrastructure • IN2P3 forest in production • 9 labs in production (8 sites) • 4 labs should join shortly (by the end of June) • No problem so far… • Forest management not to heavy • Domain management delegated to labs • Move of LAL domain to IN2P3 forest delayed • Planned this summer • Challenge : transparent to users, preserving GPOs • Back to NT4 domain and reupgrade to ActiveDirectory (2003) LAL Site Report - HEPix - Edinburgh 2004
Virus • Not too much affected • Ports attacked filtered in border router • 2 PCs infected by Sasser outside LAL • Checking is time consuming… • Proactive patch installation is critical • Main tool : SMS • Rapid deployment of fixes : 2 hours for 90% of running PCs • Integrating SMS and SUS : SUS feature pack • SUS integrated with SMS inventory and logging capabilites LAL Site Report - HEPix - Edinburgh 2004
Resources Monitoring • Operation Control Center (Tableau de bord) • LAL Solution based on Nagios and RRDTools • HTML front end with a service oriented view • Grouping / aggregation of Nagios status • Alerts and notifications capabilities • Proactive/corrective actions associated with alerts • Performance data with RRDTools (former MRTG) • Services monitored by plugins • Almost everything can be monitored • No platform / OS dependency • Remote execution of plugins through SSH LAL Site Report - HEPix - Edinburgh 2004
Operation Control Center LAL Site Report - HEPix - Edinburgh 2004
Operation Control Center LAL Site Report - HEPix - Edinburgh 2004
GRID • Strong involvement from LAL in EDG / LCG / EGEE • 3 FTE during EDG/LCG (including EDG Integration Team leader) • EGEE : 1 of the official french site, 2 FTE funded • 1 middleware, 1 application deployment • Fabric update : 26 CPUs to be added shortly (total : 36) • Mainly Xeon 3 Ghz and Opteron 2,2 Ghz • Involved in a P2P computing project (XtremWeb) • Partnership with LRI (Orsay, computing research) • 1 FTE (LAL engineer preparing a PHD) • 30 CPUs (mainly Xeon 3 Ghz and Opteron 2,2 Ghz) • LAL interest : use of P2P ressource as a fabric via OGSA LAL Site Report - HEPix - Edinburgh 2004
Miscellaneous Projects • Unattended Linux installation server • Currently based on Kickstart for initial installation • Still planning to investigate EDG WP4 Quattor • Initial installation and updates • Servers and desktops • In conjunction with EGEE involvement (P. Ponianski may join LAL) • Automatic visitor registration and DHCP quarantine • Coupled with a rework of our computers database • Expected one fellow but he resigned after 2 weeks… LAL Site Report - HEPix - Edinburgh 2004