Hi All,

I tried using hwrng as source for VM I got below error.

qemu/backends/rng-random.c:44:entropy_available: assertion failed: (len != -1)

Please help to understand it more.

Thanks,
~Rohit

On Mon, Jul 3, 2017 at 8:22 PM, TranceWorldLogic . <tranceworldlogic@gmail.com> wrote:
Hi Yaniv,

I tried looking into rng direction and I found below stats.
I am not familiar with rng device but it look to me /dev/urandom giving me better option.
But I am unaware how can I use urandom device in ovirt.

RANDOM DEVICE ======>
cat /dev/random | rngtest -c 1000
rngtest 5
Copyright (c) 2004 by Henrique de Moraes Holschuh
This is free software; see the source for copying conditions.  There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

rngtest: starting FIPS tests...
rngtest: bits received from input: 20000032
rngtest: FIPS 140-2 successes: 1000
rngtest: FIPS 140-2 failures: 0
rngtest: FIPS 140-2(2001-10-10) Monobit: 0
rngtest: FIPS 140-2(2001-10-10) Poker: 0
rngtest: FIPS 140-2(2001-10-10) Runs: 0
rngtest: FIPS 140-2(2001-10-10) Long run: 0
rngtest: FIPS 140-2(2001-10-10) Continuous run: 0
rngtest: input channel speed: (min=3.594; avg=4.813; max=5.968)Mibits/s
rngtest: FIPS tests speed: (min=93.958; avg=129.073; max=157.632)Mibits/s
rngtest: Program run time: 4111375 microseconds

URANDOM DEVICE ======>
cat /dev/urandom | rngtest -c 1000
rngtest 5
Copyright (c) 2004 by Henrique de Moraes Holschuh
This is free software; see the source for copying conditions.  There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

rngtest: starting FIPS tests...
rngtest: bits received from input: 20000032
rngtest: FIPS 140-2 successes: 1000
rngtest: FIPS 140-2 failures: 0
rngtest: FIPS 140-2(2001-10-10) Monobit: 0
rngtest: FIPS 140-2(2001-10-10) Poker: 0
rngtest: FIPS 140-2(2001-10-10) Runs: 0
rngtest: FIPS 140-2(2001-10-10) Long run: 0
rngtest: FIPS 140-2(2001-10-10) Continuous run: 0
rngtest: input channel speed: (min=1.035; avg=17.311; max=18.626)Gibits/s
rngtest: FIPS tests speed: (min=119.959; avg=161.107; max=164.427)Mibits/s
rngtest: Program run time: 120154 microseconds

Thanks,
~Rohit



On Fri, Jun 30, 2017 at 7:42 PM, TranceWorldLogic . <tranceworldlogic@gmail.com> wrote:
Your understanding is correct issue only due to encryption/decryption process but not got idea why not word.
I found that in centos 7 we not have install rng-tools.

Do it required to install for random generator ?

I have changes nothing I just increase number of queues in vnet. It diffenetly increase throughput and create multiple softIRQs in VM.
But for normal traffic this all things are not required it gives 10G throughput.

Thanks,
~Rohit

On Fri, Jun 30, 2017 at 7:08 PM, Yaniv Kaul <ykaul@redhat.com> wrote:


On Fri, Jun 30, 2017 at 4:14 PM, TranceWorldLogic . <tranceworldlogic@gmail.com> wrote:
Hi Yaniv,

I have enabled random generator in cluster and also in VM.
But still not see any improvement in throughput.

lsmod | grep -i virtio
virtio_rng             13019  0

Are you sure it's being used? What is the qemu command line (do you see the device in the guest?)
 
virtio_balloon         13834  0
virtio_console         28115  2
virtio_blk             18156  4
virtio_scsi            18361  0
virtio_net             28024  0
virtio_pci             22913  0
virtio_ring            21524  7 virtio_blk,virtio_net,virtio_pci,virtio_rng,virtio_balloon,virtio_console,virtio_scsi
virtio                 15008  7 virtio_blk,virtio_net,virtio_pci,virtio_rng,virtio_balloon,virtio_console,virtio_scsi

Would please check do I missing some virtio module ?

One more finding, if I set queue property in vnic profile then I got good throughput.

Interesting - I had assumed the bottleneck would be the encryption/decryption process, not the network. What do you set exactly? Does it matter in non-encrypted traffic as well? Are the packets (and the whole communication) large or small (i.e, would jumbo frames help) ?
 Y.


Thanks,
~Rohit


On Fri, Jun 30, 2017 at 12:11 AM, Yaniv Kaul <ykaul@redhat.com> wrote:


On Thu, Jun 29, 2017 at 4:02 PM, TranceWorldLogic . <tranceworldlogic@gmail.com> wrote:
Got it, just I need to do modprobe to add virtio-rng driver.
I will try with this option.

Make sure it is checked on the cluster.
Y. 

Thanks for your help,
~Rohit

On Thu, Jun 29, 2017 at 6:20 PM, TranceWorldLogic . <tranceworldlogic@gmail.com> wrote:
Hi,

I am using host as Centos 7.3 and guest also centos 7.3
it have 3.10 kernel version.

But I not see virtio-rng in guest VM. Is this module come with kernel or separately I have to install ?

Thanks,
~Rohit

On Thu, Jun 29, 2017 at 5:28 PM, Yaniv Kaul <ykaul@redhat.com> wrote:


On Thu, Jun 29, 2017 at 10:07 AM, TranceWorldLogic . <tranceworldlogic@gmail.com> wrote:
Hi,

To increase network throughput we have changed txqueuelen of network device and bridge manually. And observed  improved throughput.

Interesting, as I've read the default (1000) should be good enough for 10g, for example.

Are you actually seeing errors on the interface (overruns) and such?
 

But in ovirt I not see any option to increase txqueuelen.

Perhaps use ifup-local script to set it when the interface goes up?
 
Can someone suggest me what will be the right way to increase throughput ?

Note: I am trying to increase throughput for  ipsec packets.

For ipsec, probably best to ensure virtio-rng is enabled.
Y.
 

Thanks,
~Rohit

_______________________________________________
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users