openstack/nova
Code Issues Proposed changes

Always pass 'NUMACell.siblings' to _pack_instance_onto_cores'

Browse Source 
Now that we're populating the aforementioned attribute with cores that
don't have siblings, we don't need to continue creating imitation lists
of "siblings".

Change-Id: Ia371684dd9f821f214c6e023845875021848ca0a
Related-Bug: #1744965
This commit is contained in:
Stephen Finucane
2018-01-23 16:09:10 +00:00
parent f200a4ad26
commit 5e899e8887
1 changed files with 15 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
nova/virt/hardware.py
View File

@@ -843,6 +843,15 @@ def _pack_instance_onto_cores(available_siblings,
instance_cell.cpuset, instance_cell.cpuset,
num_cpu_reserved=num_cpu_reserved) num_cpu_reserved=num_cpu_reserved)
else: # REQUIRE, PREFER (explicit, implicit) else: # REQUIRE, PREFER (explicit, implicit)
if (instance_cell.cpu_thread_policy ==
fields.CPUThreadAllocationPolicy.REQUIRE):
# make sure we actually have some siblings to play with
if threads_per_core <= 1:
LOG.info("Host does not support hyperthreading or "
"hyperthreading is disabled, but 'require' "
"threads policy was requested.")
return
# NOTE(ndipanov): We iterate over the sibling sets in descending order # NOTE(ndipanov): We iterate over the sibling sets in descending order
# of cores that can be packed. This is an attempt to evenly distribute # of cores that can be packed. This is an attempt to evenly distribute
# instances among physical cores # instances among physical cores
@@ -941,27 +950,12 @@ def _numa_fit_instance_cell_with_pinning(host_cell, instance_cell,
'total': host_cell.memory}) 'total': host_cell.memory})
return return
if host_cell.siblings: # Try to pack the instance cell onto cores
LOG.debug('Using thread siblings for packing') # TODO(stephenfin): We should just pass host_cell to this function
# Try to pack the instance cell onto cores numa_cell = _pack_instance_onto_cores(
numa_cell = _pack_instance_onto_cores( host_cell.free_siblings, instance_cell, host_cell.id,
host_cell.free_siblings, instance_cell, host_cell.id, max(map(len, host_cell.siblings)),
max(map(len, host_cell.siblings)), num_cpu_reserved=num_cpu_reserved)
num_cpu_reserved=num_cpu_reserved)
else:
if (instance_cell.cpu_thread_policy ==
fields.CPUThreadAllocationPolicy.REQUIRE):
LOG.info("Host does not support hyperthreading or "
"hyperthreading is disabled, but 'require' "
"threads policy was requested.")
return
# Straightforward to pin to available cpus when there is no
# hyperthreading on the host
free_cpus = [set([cpu]) for cpu in host_cell.free_cpus]
numa_cell = _pack_instance_onto_cores(
free_cpus, instance_cell, host_cell.id,
num_cpu_reserved=num_cpu_reserved)
if not numa_cell: if not numa_cell:
LOG.debug('Failed to map instance cell CPUs to host cell CPUs') LOG.debug('Failed to map instance cell CPUs to host cell CPUs')