GK Infrastructure
Gatekeeper Infrastructure
This page was last updated on May 17, 2016.
The nodes for STAR's grid-related activities at BNL are as follows:
Color coding
- Black: in production (please, do NOT modify without prior warning)
- Green: machine was setup for testing particular component or setup
- Red : status unknown
- Blue: available for upgrade upon approval
Grid Machine | Usage | Notes | Hardware Make and Model | OS version, default gcc version |
Hardware arrangement | OSG base | Condor |
---|---|---|---|---|---|---|---|
stargrid01 | FROM BNL, submit grid jobs from this node | Dell PowerEdge 2950 dual quad-core Xeon E5440 (2.83 GHz/ 1.333 GHz FSB), 16 GB RAM |
RHEL Client 5.11, gcc 4.3.2 |
6 x 1TB SATA2: 1GB /boot (/dev/md0) is RAID 1 across all six drives There are 3 RAID 1 arrays using pairs of disks (eg. /dev/sda2 and /dev/sdb2 are one array). The various local mount points and swap space are logical volumes scattered across these RAIDed pairs. There are 2.68 TB of unassigned space in the current configuration. |
OSG 3.2.25 Client software stack for job submission | 8.2.8-1.4 (part of OSG install -- only for grid submission, not part of RACF condor) | |
stargrid02 | file transfer (gridftp) server | Attention: on stargrid02, the mappings *formerly* were all grid mappings (i.e. to VO group accounts: osgstar, engage, ligo, etc...) On May 17, 2016, this was changed to map STAR VO users to individual user accounts (matching the behaviour of stargrid03 and stargrid04). This behavior may be changed back. (TBD) Former STAR-BNL site gatekeeper |
Dell PowerEdge 2950 dual quad-core Xeon E5440 (2.83 GHz/ 1.333 GHz FSB), 16 GB RAM |
RHEL Client 5.11, gcc 4.3.2 |
6 x 1TB SATA2: Configured the same as stargrid01 above NIC 2 x 1Gb/s (one in use for RACF IPMI/remote administration on private network) |
OSG CE 3.1.23 |
7.6.10 (RCF RPM), NON-FUNCTIONAL (non-working configuration) |
stargrid03 | file transfer (gridftp) server | To transfer using STAR individual user mappings, please use this node or stargrid04 | Dell PowerEdge 2950 dual quad-core Xeon E5440 (2.83 GHz/ 1.333 GHz FSB), 16 GB RAM |
RHEL Client 5.11, gcc 4.3.2 |
6 x 1TB SATA2: Configured the same as stargrid01 above NIC 2 x 1Gb/s (one in use for RACF IPMI/remote administration on private network) |
OSG CE 3.1.18 | 7.6.10 (RCF RPM), NON-FUNCTIONAL (non-working configuration) |
stargrid04 | file transfer (gridftp) server | To transfer using STAR individual user mappings, please use this node or stargrid03 | Dell PowerEdge 2950 dual quad-core Xeon E5440 (2.83 GHz/ 1.333 GHz FSB), 16 GB RAM |
RHEL Client 5.11, gcc 4.3.2 |
6 x 1TB SATA2: Configured the same as stargrid01 above NIC 2 x 1Gb/s (one in use for RACF IPMI/remote administration on private network) |
OSG CE 3.1.23 | 7.6.10 (RCF RPM), NON-FUNCTIONAL (non-working configuration) |
stargrid0[234] are using the VDT-supplied gums client (version 1.2.16).
stargrid02 has a local hack in condor.pm to adjust the condor parameters for STAR users with local accounts.
All nodes have GLOBUS_TCP_PORT_RANGE=20000,30000 and matching firewall conduits for Globus and other dynamic grid service ports.
- General upgrade history (generally obsolete, but some valuable tidbits)
- stargridXX (generally obsolete, but some valuable tidbits)
- stargrid02 (generally obsolete, but some valuable tidbits)
- stargrid04 (generally obsolete, but some valuable tidbits)
- Printer-friendly version
- Login or register to post comments