LSF Version 7.3 - Platform LSF Configuration Reference

•
MEMLIMIT : Corresponds to JLIMIT_RSS; Resident pages above the limit become prime
swap candidates
•
DATALIMIT : Corresponds to LIMIT_DATA; malloc() calls in the job fail with errno
set to ENOMEM
•
CPULIMIT: Corresponds to JLIMIT_CPU; IRIX sends SIGXCPU signal to job, then after
the grace period expires, sends SIGINT, SIGTERM, and SIGKILL
•
FILELIMIT: No corresponding IRIX limit; use process limit RLIMIT_FSIZE
•
STACKLIMIT : No corresponding IRIX limit; use process limit RLIMIT_STACK
•
CORELIMIT: No corresponding IRIX limit; use process limit RLIMIT_CORE
•
SWAPLIMIT: Corresponds to JLIMIT_VMEM; use process limit RLIMIT_VMEM
Increasing the default MEMLIMIT for ULDB
In some pre-defined LSF queues, such as normal, the default MEMLIMIT is set to 5000 (5
MB). However, if ULDB is enabled (LSF_ULDB_DOMAIN is defined) the MEMLIMIT
should be set greater than 8000 in lsb.queues.
Default
Not defined
LSF_UNIT_FOR_LIMITS
Syntax
LSF_UNIT_FOR_LIMITS=unit
Description
Enables scaling of large units in resource usage limits.
When set, LSF_UNIT_FOR_LIMITS applies cluster-wide to limits at the job-level (bsub),
queue-level (lsb.queues), and application level (lsb.applications).
The limit unit specified by LSF_UNIT_FOR_LIMITS also applies to limits modified with
bmod, and the display of resource usage limits in query commands (bacct, bapp, bhist,
bhosts, bjobs, bqueues, lsload, and lshosts).
Important:
Before changing the units of your resource usage limits, you
should completely drain the cluster of all workload. There should
be no running, pending, or finished jobs in the system.
In a MultiCluster environment, you should configure the same unit for all clusters.
Example
A job is submitted with bsub -M 100 and
LSF_UNIT_FOR_LIMITS=MB; the memory limit for the job is 100 MB
rather than the default 100 KB.
lsf.conf
498 Platform LSF Configuration Reference