Difference between revisions of "RAL Memory Limits"
From GridPP Wiki
(Created page with "All jobs on the RAL batch system are run in distinct cgroups, using the following cgroup subsystems: cpu, cpuacct, memory, freezer and blkio. On most worker nodes there are no...") |
|||
| Line 1: | Line 1: | ||
| − | All jobs on the RAL batch system are run in distinct cgroups, using the following cgroup subsystems: cpu, cpuacct, memory, freezer and blkio. On most worker nodes there are no memory limits applied using cgroups. However, on one tranche of worker nodes, corresponding to around 2000 cores, we have soft memory limits applied. In this case the cgroup attribute ''memory.soft_limit_in_bytes'' for each job is set to the amount of memory requested by the job. | + | All jobs on the RAL batch system are run in distinct cgroups, using the following cgroup subsystems: cpu, cpuacct, memory, freezer and blkio. On most worker nodes there are no memory limits applied using cgroups. However, on one tranche of worker nodes, corresponding to around 2000 cores, we have soft memory limits applied. In this case the cgroup attribute ''memory.soft_limit_in_bytes'' for each job is set to the amount of memory requested by the job. Jobs are allowed to exceed this memory limit if there is free memory available on the system. Only when there is contention between other processes for physical memory will the system force physical memory into swap and push the physical memory used towards the assigned limit. |
Revision as of 13:34, 13 October 2014
All jobs on the RAL batch system are run in distinct cgroups, using the following cgroup subsystems: cpu, cpuacct, memory, freezer and blkio. On most worker nodes there are no memory limits applied using cgroups. However, on one tranche of worker nodes, corresponding to around 2000 cores, we have soft memory limits applied. In this case the cgroup attribute memory.soft_limit_in_bytes for each job is set to the amount of memory requested by the job. Jobs are allowed to exceed this memory limit if there is free memory available on the system. Only when there is contention between other processes for physical memory will the system force physical memory into swap and push the physical memory used towards the assigned limit.
