[dpdk-users] rte_eal_init fails with --socket-mem set to more than 192MB per socket

[Wiles, Keith]

Regards,
Keith

> On Aug 31, 2016, at 2:15 PM, Sarthak Ray <sarthak_ray at outlook.com> wrote:
> 
> Hi,
> 
> I am using dpdk-2.1.0 and I am not able to reserve memory beyond 192MB per socket by using --socket-mem option. I see below error logs, though my system has enough free memory.
> 
> EAL: Not enough memory available on socket 0! Requested: 256MB, available: 192MB
> PANIC in rte_eal_init():
\Most of the time this means that contiguous memory is not available and you have fragmented huge pages. The normal fix is to make sure you allocate the huge pages early in boot up, which to me means making sure you have the /etc/sysctl.conf file setup with the number of huge pages.

vm.nr_hugepages=NNN

> 
> # numactl -H
> available: 2 nodes (0-1)
> node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
> node 0 size: 65170 MB
> node 0 free: 47433 MB
> node 1 cpus: 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
> node 1 size: 65536 MB
> node 1 free: 49999 MB
> node distances:
> node   0   1
>  0:  10  21
>  1:  21  10
> 
> Is there any max limit for reserving memory per socket? If yes, then please suggest me how to increase that limit.
> 
> Thanks in advance,
> Sarthak