additional cgroup testing, implementation of cpu-shares, and introduction of blkio limits. #1082
Conversation
|
LGTM |
cdoern
force-pushed
the
cgroup
branch
2 times, most recently
from
e303216 to
09c91eb
Compare
|
repushed with cpuset variables all equal to "0", testing just one node. |
|
/hold working on an io implementation on this branch. I found an issue in godbus that i think needs to be patched first @giuseppe PTAL at godbus/dbus#328 |
cdoern
changed the title
|
testing will fail until godbus/dbus#329 is merged. |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: cdoern, giuseppe, rhatdan The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
seems like the tests pass even when the value of the property is |
cdoern
force-pushed
the
cgroup
branch
6 times, most recently
from
13a3dae to
b73f940
Compare
added support cpu-shares/weight and most Blkio throttle/weight devices. podman currently support ThrottleReadBps devices so I added those, writeBps, and weight devices. write/read iops seems to be lesser supported but will see if that is a possibility in the future. this involved some rewiring in godbus and some questions in systemd. Seems like systemd is pretty rigorous with their device paths, requiring a full path to either char or block as well as the major:minor. This was difficult to debug and will most likely lead to design limitations in the future. Signed-off-by: Charlie Doern <cdoern@redhat.com>
|
/lgtm |
|
/hold cancel |
added some tests for more resource limits and put translation of cpu-weight in there. Podman and runc both only use cpu-shares but systemd has replaced CPUShares with CPUWeight in more recent versions, they are the same thing it seems.
Signed-off-by: Charlie Doern cdoern@redhat.com