When physical interface is hosted on MPC without enhanced queuing capabilities (MPCE Type 1 3D, MPCE Type 2 3D, MPC4E 3D) has considerable amount of logical interfaces configured and 'per-unit-scheduler' interface configuration is present, router may report messages similar to:

COS-HALP: IFL et-2/1/0.730 (ifindex 1464) failed to get scheduling(L2) resource.
Instead it will use parent interface shared queues.
PFE_ERROR_NOT_FOUND: cos_halp_ifl does not exist for ifl:1464
COS(cos_ifl_walk:7614): walker failed IFL:1464 Ifd 158

'per-unit-scheduler' feature allows users to enable scheduling and shaping at logical interfaces on non-queuing MPCs and hence providing customized traffic shaping services which are independent of other logical interfaces on a given physical interface. It's configured as:

interfaces {
  <physical interface name> {
    per-unit-scheduler;  
  }
}

This feature supported limited scale up to 62 logical interfaces per interface and up to 256 logical interfaces per PFE complex. When more logical interfaces are configured on top of physical interface, logical interface will fail to get requested queues and will be attached to physical interface common queue. If such event happens, the syslog warning message as above will be generated.

The log message above is expected when more that 62 logical interfaces are configured to use this feature per physical interface. To solve this issue on non-queuing FPCs, it's recommended to reduce the scale to be bellow mentioned one. If scale above 62 logical interfaces per physical interface is required, it's recommended to use queuing MPC (MPCE Type 2 3D Q, MPCE Type 2 3D EQ, MPC2E NG PQ & Flex Q, MPC2E HQoS etc)