Date   

Re: Feature Project proposal

Tapio Tallgren
 

Not really. Perhaps the issues that you should consider are

- This project will make available a proprietary tool for testing Akraino blueprints. Akraino developers can use the tool freely and can write their own test cases.
I am ok with this. You could ask for an SLA since one cannot fix the bugs oneself :-)

- One might wonder how many projects will use this tool, but they actually had two blueprints volunteering, so ok from my pov.

- Support for the 5G protocol stack is work in progress. Just a comment, no issues with this.

________________________________________
From: Andrew Wilkinson <andrew.wilkinson@...>
Sent: Tuesday, June 16, 2020 17:22
To: Tallgren, Tapio (Nokia - FI/Espoo); technical-discuss@...
Subject: FW: Feature Project proposal

Hi Tapio and those on the last community call,

As I wasn’t able to attend the call could you summarize any input on the new proposal – were there any items related to adopting the FP identified by the community?

Andrew

From: process@... <process@...> On Behalf Of Samir Chatterjee via lists.akraino.org
Sent: Thursday, June 11, 2020 07:25
To: process@...
Cc: Tina Tsou <tina.tsou@...>; 'Kannan Subbiah (kansubbi) (kansubbi@...)' <kansubbi@...>; Pradnesh Dange <pradnesh.dange@...>
Subject: [Akraino Process Sub-Committee] Feature Project proposal

Folks

I would like an opportunity to present the proposed Call Flow Feature Project proposal. A brief presentation of the proposal is at https://wiki.akraino.org/display/AK/End-to-End+Call+Flow+Validation+of+BP?src=contextnavpagetreemode<https://protect2.fireeye.com/v1/url?k=6a298840-34894a05-6a29c8db-86b568293eb5-247d2dd35b45a874&q=1&e=8d103582-8ab1-44c9-873f-208f42ab239a&u=https%3A%2F%2Fwiki.akraino.org%2Fdisplay%2FAK%2FEnd-to-End%2BCall%2BFlow%2BValidation%2Bof%2BBP%3Fsrc%3Dcontextnavpagetreemode>

We have presented the proposal today at Community Subcommittee, a recording of that should be available for review. Please let me know when we can have an opportunity to present this for your consideration.

Regards,
Samir


::Disclaimer:: Information contained and transmitted by this E-mail communication including any attachment is intended solely for the exclusive use of the addressee(s) and may contain information that is privileged, confidential or exempt from disclosure. Access to this e-mail and/or to the attachment by anyone else is unauthorised. If you are not the intended recipient, an agent of the intended recipient or a person responsible for delivering the information to the named recipient, you are notified that any use, distribution, transmission, printing, copying or dissemination of this information in any way or in any manner is strictly prohibited. If you are not the intended recipient of this mail kindly destroy all copies of this message and any attachments and inform the sender.

::WARNING:: Although this communication is believed to be free of any virus which might affect any computer or IT system into which they are opened, the recipient should, by all means, check this message and any attachments for the presence of any virus. The sender and sender organisation accept no liability for any damage caused or the potential thereof by any virus transmitted by this message. Thank you !


FW: Feature Project proposal

Andrew Wilkinson
 

Hi Tapio and those on the last community call,

 

As I wasn’t able to attend the call could you summarize any input on the new proposal – were there any items related to adopting the FP  identified by the community?

 

Andrew

 

From: process@... <process@...> On Behalf Of Samir Chatterjee via lists.akraino.org
Sent: Thursday, June 11, 2020 07:25
To: process@...
Cc: Tina Tsou <tina.tsou@...>; 'Kannan Subbiah (kansubbi) (kansubbi@...)' <kansubbi@...>; Pradnesh Dange <pradnesh.dange@...>
Subject: [Akraino Process Sub-Committee] Feature Project proposal

 

Folks

 

I would like an opportunity to present the proposed Call Flow Feature Project proposal. A brief presentation of the proposal is at https://wiki.akraino.org/display/AK/End-to-End+Call+Flow+Validation+of+BP?src=contextnavpagetreemode

 

We have presented the proposal today at Community Subcommittee, a recording of that should be available for review. Please let me know when we can have an opportunity to present this for your consideration.

 

Regards,

Samir



::Disclaimer:: Information contained and transmitted by this E-mail communication including any attachment is intended solely for the exclusive use of the addressee(s) and may contain information that is privileged, confidential or exempt from disclosure. Access to this e-mail and/or to the attachment by anyone else is unauthorised. If you are not the intended recipient, an agent of the intended recipient or a person responsible for delivering the information to the named recipient, you are notified that any use, distribution, transmission, printing, copying or dissemination of this information in any way or in any manner is strictly prohibited. If you are not the intended recipient of this mail kindly destroy all copies of this message and any attachments and inform the sender.

::WARNING:: Although this communication is believed to be free of any virus which might affect any computer or IT system into which they are opened, the recipient should, by all means, check this message and any attachments for the presence of any virus. The sender and sender organisation accept no liability for any damage caused or the potential thereof by any virus transmitted by this message. Thank you !


Upcoming Event: Akraino Technical Community Call (Weekly) - Thu, 06/11/2020 1:00pm-2:00pm #cal-reminder

technical-discuss@lists.akraino.org Calendar <technical-discuss@...>
 

Reminder: Akraino Technical Community Call (Weekly)

When: Thursday, 11 June 2020, 1:00pm to 2:00pm, (GMT+00:00) UTC

Where:https://zoom.us/j/919148693

View Event

Organizer: technical-discuss@...

Description:

Akraino Technical Community Call: TSC updates to technical community and deeper dives into topics as applicable. Meeting content posted to Technical Community Wiki.
Meeting Lead: Kandan Kathirvel, Akraino TSC Chair


Akraino Edge Stack is inviting you to a scheduled Zoom meeting.
Join from PC, Mac, Linux, iOS or Android:
https://zoom.us/j/919148693
Or iPhone one-tap : US: +16699006833,,919148693# or +16465588656,,919148693#
Or Telephone: Dial(for higher quality, dial a number based on your current location):
US: +1 669 900 6833 or +1 646 558 8656 or +1 877 369 0926 (Toll Free) or +1 855 880 1246 (Toll Free)
Meeting ID: 919 148 693
International numbers available:
https://zoom.us/u/adnlim1pfM


Upcoming Event: Akraino Technical Community Call (Weekly) - Thu, 06/11/2020 1:00pm-2:00pm #cal-reminder

technical-discuss@lists.akraino.org Calendar <technical-discuss@...>
 

Reminder: Akraino Technical Community Call (Weekly)

When: Thursday, 11 June 2020, 1:00pm to 2:00pm, (GMT+00:00) UTC

Where:https://zoom.us/j/919148693

View Event

Organizer: technical-discuss@...

Description:

Akraino Technical Community Call: TSC updates to technical community and deeper dives into topics as applicable. Meeting content posted to Technical Community Wiki.
Meeting Lead: Kandan Kathirvel, Akraino TSC Chair


Akraino Edge Stack is inviting you to a scheduled Zoom meeting.
Join from PC, Mac, Linux, iOS or Android:
https://zoom.us/j/919148693
Or iPhone one-tap : US: +16699006833,,919148693# or +16465588656,,919148693#
Or Telephone: Dial(for higher quality, dial a number based on your current location):
US: +1 669 900 6833 or +1 646 558 8656 or +1 877 369 0926 (Toll Free) or +1 855 880 1246 (Toll Free)
Meeting ID: 919 148 693
International numbers available:
https://zoom.us/u/adnlim1pfM


Akraino Community call on Thursday June 11th at 6 am PST: New feature project presentation!

Tapio Tallgren
 

Hello all,

We have a new feature project presentation on Thursday: Call Flow Validation Feature project.

Link to the project page is https://wiki.akraino.org/display/AK/End-to-End+Call+Flow+Validation+of+BP

and coordinates to the call are:

Akraino Edge Stack 1 is inviting you to a scheduled Zoom meeting.

Topic: Akraino Technical Community Call (Weekly)
Time: Jun 11, 2020 06:00 AM Pacific Time (US and Canada)
Every week on Thu, until Aug 20, 2020, 11 occurrence(s)
Jun 11, 2020 06:00 AM
Jun 18, 2020 06:00 AM
Jun 25, 2020 06:00 AM
Jul 2, 2020 06:00 AM
Jul 9, 2020 06:00 AM
Jul 16, 2020 06:00 AM
Jul 23, 2020 06:00 AM
Jul 30, 2020 06:00 AM
Aug 6, 2020 06:00 AM
Aug 13, 2020 06:00 AM
Aug 20, 2020 06:00 AM
Please download and import the following iCalendar (.ics) files to your calendar system.
Weekly: https://zoom.us/meeting/tJUkdu2gqDIpnBNDikhC6gIzeAfMEExPDg/ics?icsToken=98tyKuCppzouEtKcs139Z60tA97qb9_Ikn9shIdUlxfGVSd6eAbXMM9nF6pPNN-B

Join Zoom Meeting
https://zoom.us/j/919148693

Meeting ID: 919 148 693
One tap mobile
+13017158592,,919148693# US (Germantown)
+13126266799,,919148693# US (Chicago)

Dial by your location
+1 301 715 8592 US (Germantown)
+1 312 626 6799 US (Chicago)
+1 346 248 7799 US (Houston)
+1 646 558 8656 US (New York)
+1 669 900 6833 US (San Jose)
+1 253 215 8782 US (Tacoma)
877 369 0926 US Toll-free
855 880 1246 US Toll-free
Meeting ID: 919 148 693
Find your local number: https://zoom.us/u/adnlim1pfM


Call Flow Validation Feature project

Pradnesh Dange <pradnesh.dange@...>
 

Hello TSC , Process and Technical Committee members,
We have put up the End to End Call Flow validation feature project proposal at https://wiki.akraino.org/display/AK/End-to-End+Call+Flow+Validation+of+BP?src=contextnavpagetreemode

Requesting you to let us know review slots accordingly.


Regards,

Pradnesh 

Rebaca Technologies 

408-623-8420










::Disclaimer:: Information contained and transmitted by this E-mail communication including any attachment is intended solely for the exclusive use of the addressee(s) and may contain information that is privileged, confidential or exempt from disclosure. Access to this e-mail and/or to the attachment by anyone else is unauthorised. If you are not the intended recipient, an agent of the intended recipient or a person responsible for delivering the information to the named recipient, you are notified that any use, distribution, transmission, printing, copying or dissemination of this information in any way or in any manner is strictly prohibited. If you are not the intended recipient of this mail kindly destroy all copies of this message and any attachments and inform the sender.

::WARNING:: Although this communication is believed to be free of any virus which might affect any computer or IT system into which they are opened, the recipient should, by all means, check this message and any attachments for the presence of any virus. The sender and sender organisation accept no liability for any damage caused or the potential thereof by any virus transmitted by this message. Thank you !


Re: About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@...>
 

Dear Paul,

Thank you so much for the detailed historical review. Now I think I've got good idea.

As we are using Adlink ALPS-7400Appliance, which has built-in HA, with redundant disk in its 2 MH (managed nodes),
which mapped the disk to its 4 CH (compute nodes), where we'll use 3 as controller and 1 as worker.
We believe that we can run Ceph and let Ceph do the redundancy.

Sincerely,
Frank

-----Original Message-----
From: CARVER, PAUL [mailto:pc2929@...]
Sent: Monday, June 8, 2020 8:52 PM
To: 蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@...>; technical-discuss@...
Cc: 詹子儀 Andy Chan <andychan@...>; Sean Xie <sean.xie@...>
Subject: RE: About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

I couldn’t remember where those numbers came from so I asked a few people who were involved when we started the project. I know we never did any testing to try to find the absolute minimum disk required, so it looks like we just documented the actual disk sizes of the hardware we used for testing the system as a whole. We have tested REC on Dell and HP machines with far more disks than are required because we were able to borrow the servers from other projects rather than buying specifically for REC, but the numbers you’re asking about are derived from our “reference platform”, i.e. the specific cluster that Jenkins automatically installs the REC onto in our Continuous Deployment pipeline as described in https://wiki.akraino.org/display/AK/Radio+Edge+Cloud+Validation+Lab <https://wiki.akraino.org/display/AK/Radio+Edge+Cloud+Validation+Lab>



In that cluster we are using the Nokia OpenEdge single height, half width server blade which is equipped with two 480GB M.2 boards (i.e. something that looks similar to a RAM module plugged into the motherboard) and two 960GB 2.5 inch SSDs in front mounted slots. Since the minimum cluster size is three controllers and zero workers the total disk capacity of a minimum size cluster on this specific hardware is 6 of the M.2 boards and 6 of the 2.5 inch disks. That’s where the numbers on that wiki page are coming from.



In actuality, the specific cluster in our CD cluster is using all 5 blades in a single chassis for a total of 10 of the 480GB M.2 boards and 10 of the 960GB 2.5 inch drives, but one of the M.2 boards in each server is unused. We had considered doing RAID1 on the pair of M.2 boards but these servers do not support hardware RAID and we decided that adding software RAID was not a priority since our target deployment has three controllers and two workers. All of the 2.5 inch drives in the cluster are managed by Ceph. In order to use Ceph in REC it is highly recommended to give two physical disks per server to Ceph, but the exact size of the disks isn’t very important.



We have other REC clusters in various labs deployed with 4 or 3 servers, so it certainly isn’t necessary to have 10 of each drive type.



From: 蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@...>
Sent: Thursday, June 4, 2020 10:24
To: CARVER, PAUL <pc2929@...>; technical-discuss@...
Cc: 詹子儀 Andy Chan <andychan@...>; Sean Xie <sean.xie@...>
Subject: RE: About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)



Dear Paul,



Thanks very much for the clarity. It makes me get better understanding of the underlying design considerations.

For the kernel and latency issue, yes, some benchmark is needed. Yet, for us,

We plan to try to make sure functionality is OK, and later, rather than sticking with

Kernel 3.10 and try tuning performance, we shall fix the driver issues and upgrade to kernel4.14.

So, as deployment unit of 5 servers in a 3U chassis has been a convenient size where 3 of them are controllers

and the other 2 servers are not controllers, I trust that “Total SSD-based OS Storage: 2.8 TB (6 x 480GB SSDs)” specified in

the “Hardware Requirements” is an empirical value also based on 5 servers?

Thanks again,

Frank



From: CARVER, PAUL [mailto:pc2929@...]
Sent: Thursday, June 4, 2020 10:02 PM
To: 蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@... <mailto:ftsai@...> >; technical-discuss@... <mailto:technical-discuss@...>
Cc: 詹子儀 Andy Chan <andychan@... <mailto:andychan@...> >; Sean Xie <sean.xie@... <mailto:sean.xie@...> >
Subject: RE: About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)



Here are my answers. Nokia provided the original seed code for REC/TA based on previous closed source work they had done, so there might also be historical factors that I’m not aware of.



1. Yes, there are several components in the system that use a quorum based mechanism where you need more than 50% of the total components to be in communications with each other in order to differentiate between a failure vs a split brain scenario. That effectively makes the HA minimum 3 instead of 2. There could conceivably be any odd number greater than 1 for HA but all of our testing has been based on 3.



2. Yes, “node” is referring to physical servers. Our focus on REC/TA is on integration testing a full platform suitable for production deployment in edge locations where there isn’t a pre-existing general purpose cloud. The distinction between RIC (an O-RAN SC project) and REC (an Akraino project) is that RIC is purely software whereas REC is concerned with providing the cloud infrastructure hardware on top of which the RIC can be deployed.



3. For the joint work between Nokia and AT&T to create the initial release of REC/TA we selected the Nokia OpenEdge chassis based platform as the reference hardware platform for our testing. We also did some testing on Dell and HP hardware and have subsequently seen the support for Ampere’s ARM servers added, but we started with the Nokia OpenEdge which packages 5 servers into a 3U chassis. As such, the vast majority of our testing has been on clusters of 5 servers where 3 of them are controllers and the other 2 servers are not controllers. Our use case is primarily on deploying a large number of small clusters in many different edge locations rather than general purpose datacenter cloud, so the deployment unit of 5 servers in a 3U chassis has been a convenient size.



4. I don’t think we’re currently using DPDK. REC/TA had its roots in a previous Nokia closed source OpenStack deployment system, so Open vSwitch DPDK support was provided with OpenStack Neutron, but REC/TA is not intended as an OpenStack system. It is intended as a pure Kubernetes system that makes use of a few OpenStack components such as Ironic for baremetal deployment and Keystone for authentication.



5. The kernel version was selected specifically after some benchmarking of latency. It’s not a realtime kernel, but latency is a major consideration for the RIC which is the first and primary application that we designed REC to support. That’s not to say that it’s the only kernel version that will support the latency requirements, but if you use a different kernel then it would definitely be a good idea to run some latency benchmarks. I would have to dig through some old documentation to refresh my memory on what tests we ran. I know cyclictest is a popular latency benchmark but I’m pretty sure it wasn’t the only one that we compared.



From: technical-discuss@... <mailto:technical-discuss@...> <technical-discuss@... <mailto:technical-discuss@...> > On Behalf Of ??? FRANK C. D. TSAI, Ph.D.
Sent: Thursday, June 4, 2020 06:58
To: technical-discuss@... <mailto:technical-discuss@...>
Cc: 詹子儀 Andy Chan <andychan@... <mailto:andychan@...> >; Sean Xie <sean.xie@... <mailto:sean.xie@...> >
Subject: [Akraino Technical-Discuss] About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)



Dear community,





I was wondering if I can sort out some of my questions of Akraino (while I’m using Adlink Adlink H/W ALPS-7400 as the H/W platform for REC, which is a 4U network appliance with 4 computing nodes).



Per REC Installation Guide <https://urldefense.proofpoint.com/v2/url?u=https-3A__wiki.akraino.org_display_AK_REC-2BInstallation-2BGuide&d=DwMFog&c=LFYZ-o9_HUMeMTSQicvjIg&r=HBNonG828PGilNRNwXAtdg&m=mh8kvWW-UdQbhflZ-lcornEXEDdTKLNyV2mpupHKiaI&s=I0TYBghPY_usqSp5F64gTybDVrqjTMSYdEwHxzssZ2Y&e=> (*), controller nodes (3 required), worker nodes (all optional)



1. Is the purpose of 3 controller nodes for HA (high availability)?

[note: the reason I suppose so is because our past experience with StarlingX using two controller nodes for HA]



2. When we use the term “node” in the document, do we mean physical nodes? Or it can be logical nodes (like VM)?

[note; I suppose it’s physical nodes because in the same page Hardware Requirements, it reads “Minimum of 3 nodes” and if it’s for HA, we need physical node]



3. When we say “controller nodes (3 required), worker nodes (all optional)”, do we mean to let the controller nodes play dual role also as worker nodes by running applications on top of any controller node?



Or we mean to say that REC deployment must have exactly 3 controller nodes, and after the 3 controller nodes are established, we can (and we should) then gradually add worker nodes to scale out?



4. Is it a must for the NIC of a worker node to support DPDK? if no DPDK is supported, what would be the most significant impact (like some function XYZ is then not available) ?



5. Is using 4.14 is simply because 4.14 newer than 3.10? Or, it’s because some components of REC must rely on 4.14?



[note: I so inquire is because some ALPS-7400 peripheral driver is currently compatible with kernel 3.10, yet Build-237 uses kernel 4.14. After downloading the ISO (Build-237), we plan to replace the kernel modules from 4.14 to 3.10.]



Thank you very much for your guidance,





Sincerely,

Frank



(*) https://wiki.akraino.org/display/AK/REC+Installation+Guide#RECInstallationGuide-HardwareRequirements <https://urldefense.proofpoint.com/v2/url?u=https-3A__wiki.akraino.org_display_AK_REC-2BInstallation-2BGuide-23RECInstallationGuide-2DHardwareRequirements&d=DwQFog&c=LFYZ-o9_HUMeMTSQicvjIg&r=HBNonG828PGilNRNwXAtdg&m=mh8kvWW-UdQbhflZ-lcornEXEDdTKLNyV2mpupHKiaI&s=mIsl772H65jpCR_oznHVJBZbdvp-ggiHCjI9i3IozE0&e=>


Re: About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

Paul Carver
 

I couldn’t remember where those numbers came from so I asked a few people who were involved when we started the project. I know we never did any testing to try to find the absolute minimum disk required, so it looks like we just documented the actual disk sizes of the hardware we used for testing the system as a whole. We have tested REC on Dell and HP machines with far more disks than are required because we were able to borrow the servers from other projects rather than buying specifically for REC, but the numbers you’re asking about are derived from our “reference platform”, i.e. the specific cluster that Jenkins automatically installs the REC onto in our Continuous Deployment pipeline as described in https://wiki.akraino.org/display/AK/Radio+Edge+Cloud+Validation+Lab

 

In that cluster we are using the Nokia OpenEdge single height, half width server blade which is equipped with two 480GB M.2 boards (i.e. something that looks similar to a RAM module plugged into the motherboard) and two 960GB 2.5 inch SSDs in front mounted slots. Since the minimum cluster size is three controllers and zero workers the total disk capacity of a minimum size cluster on this specific hardware is 6 of the M.2 boards and 6 of the 2.5 inch disks. That’s where the numbers on that wiki page are coming from.

 

In actuality, the specific cluster in our CD cluster is using all 5 blades in a single chassis for a total of 10 of the 480GB M.2 boards and 10 of the 960GB 2.5 inch drives, but one of the M.2 boards in each server is unused. We had considered doing RAID1 on the pair of M.2 boards but these servers do not support hardware RAID and we decided that adding software RAID was not a priority since our target deployment has three controllers and two workers. All of the 2.5 inch drives in the cluster are managed by Ceph. In order to use Ceph in REC it is highly recommended to give two physical disks per server to Ceph, but the exact size of the disks isn’t very important.

 

We have other REC clusters in various labs deployed with 4 or 3 servers, so it certainly isn’t necessary to have 10 of each drive type.

 

From: 蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@...>

Sent: Thursday, June 4, 2020 10:24
To: CARVER, PAUL <pc2929@...>; technical-discuss@...
Cc: 詹子儀 Andy Chan <andychan@...>; Sean Xie <sean.xie@...>
Subject: RE: About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

 

Dear Paul,

 

Thanks very much for the clarity.  It makes me get better understanding of the underlying design considerations.

For the kernel and latency issue, yes, some benchmark is needed.  Yet, for us,

We plan to try to make sure functionality is OK, and later,  rather than sticking with

Kernel 3.10 and try tuning performance, we shall fix the driver issues and upgrade to kernel4.14.

So, as deployment unit of 5 servers in a 3U chassis has been a convenient size where 3 of them are controllers

and the other 2 servers are not controllers, I trust that “Total SSD-based OS Storage:  2.8 TB (6 x 480GB SSDs)” specified in

the “Hardware Requirements” is an empirical value also based on  5 servers?

Thanks again,

Frank

 

From: CARVER, PAUL [mailto:pc2929@...]
Sent: Thursday, June 4, 2020 10:02 PM
To:
蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@...>; technical-discuss@...
Cc:
詹子儀 Andy Chan <andychan@...>; Sean Xie <sean.xie@...>
Subject: RE: About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

 

Here are my answers. Nokia provided the original seed code for REC/TA based on previous closed source work they had done, so there might also be historical factors that I’m not aware of.

 

  1. Yes, there are several components in the system that use a quorum based mechanism where you need more than 50% of the total components to be in communications with each other in order to differentiate between a failure vs a split brain scenario. That effectively makes the HA minimum 3 instead of 2. There could conceivably be any odd number greater than 1 for HA but all of our testing has been based on 3.

 

  1. Yes, “node” is referring to physical servers. Our focus on REC/TA is on integration testing a full platform suitable for production deployment in edge locations where there isn’t a pre-existing general purpose cloud. The distinction between RIC (an O-RAN SC project) and REC (an Akraino project) is that RIC is purely software whereas REC is concerned with providing the cloud infrastructure hardware on top of which the RIC can be deployed.

 

  1. For the joint work between Nokia and AT&T to create the initial release of REC/TA we selected the Nokia OpenEdge chassis based platform as the reference hardware platform for our testing. We also did some testing on Dell and HP hardware and have subsequently seen the support for Ampere’s ARM servers added, but we started with the Nokia OpenEdge which packages 5 servers into a 3U chassis. As such, the vast majority of our testing has been on clusters of 5 servers where 3 of them are controllers and the other 2 servers are not controllers. Our use case is primarily on deploying a large number of small clusters in many different edge locations rather than general purpose datacenter cloud, so the deployment unit of 5 servers in a 3U chassis has been a convenient size.

 

  1. I don’t think we’re currently using DPDK. REC/TA had its roots in a previous Nokia closed source OpenStack deployment system, so Open vSwitch DPDK support was provided with OpenStack Neutron, but REC/TA is not intended as an OpenStack system. It is intended as a pure Kubernetes system that makes use of a few OpenStack components such as Ironic for baremetal deployment and Keystone for authentication.

 

  1. The kernel version was selected specifically after some benchmarking of latency. It’s not a realtime kernel, but latency is a major consideration for the RIC which is the first and primary application that we designed REC to support. That’s not to say that it’s the only kernel version that will support the latency requirements, but if you use a different kernel then it would definitely be a good idea to run some latency benchmarks. I would have to dig through some old documentation to refresh my memory on what tests we ran. I know cyclictest is a popular latency benchmark but I’m pretty sure it wasn’t the only one that we compared.

 

From: technical-discuss@... <technical-discuss@...> On Behalf Of ??? FRANK C. D. TSAI, Ph.D.
Sent: Thursday, June 4, 2020 06:58
To: technical-discuss@...
Cc:
詹子儀 Andy Chan <andychan@...>; Sean Xie <sean.xie@...>
Subject: [Akraino Technical-Discuss] About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

 

Dear community,

 

 

I was wondering if I can sort out some of my questions of Akraino (while I’m using Adlink Adlink H/W ALPS-7400 as the H/W platform for REC, which is a 4U network appliance with 4 computing nodes).

 

Per REC Installation Guide (*), controller nodes (3 required), worker nodes (all optional)

 

  1. Is the purpose of 3 controller nodes for HA (high availability)? 

[note: the reason I suppose so is because our past experience with StarlingX using two controller nodes for HA]

 

  1. When we use the term “node” in the document, do we mean physical nodes? Or it can be logical nodes (like VM)?

[note; I suppose it’s physical nodes because in the same page Hardware Requirements, it reads “Minimum of 3 nodes” and if it’s for HA, we need physical node]

 

  1. When we say “controller nodes (3 required), worker nodes (all optional)”, do we mean to let the controller nodes play dual role also as worker nodes by running applications on top of any controller node? 

 

Or we mean to say that REC deployment must have exactly 3 controller nodes, and after the 3 controller nodes are established, we can (and we should) then gradually add worker nodes to scale out?

 

  1. Is it a must for the NIC of a worker node to support DPDK? if no DPDK is supported, what would be the most significant impact (like some function XYZ is then not available) ?

 

  1. Is using 4.14 is simply because 4.14 newer than 3.10?  Or, it’s because some components of REC must rely on 4.14?

 

[note: I so inquire is because some ALPS-7400 peripheral driver is currently compatible with kernel 3.10, yet Build-237 uses kernel 4.14.  After downloading the ISO (Build-237), we plan to replace the kernel modules from 4.14 to 3.10.]

 

Thank you very much for your guidance,

 

 

Sincerely,

Frank

 

(*) https://wiki.akraino.org/display/AK/REC+Installation+Guide#RECInstallationGuide-HardwareRequirements

 

 


Re: About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@...>
 

Dear Paul,

 

Thanks very much for the clarity.  It makes me get better understanding of the underlying design considerations.

For the kernel and latency issue, yes, some benchmark is needed.  Yet, for us,

We plan to try to make sure functionality is OK, and later,  rather than sticking with

Kernel 3.10 and try tuning performance, we shall fix the driver issues and upgrade to kernel4.14.

So, as deployment unit of 5 servers in a 3U chassis has been a convenient size where 3 of them are controllers

and the other 2 servers are not controllers, I trust that “Total SSD-based OS Storage:  2.8 TB (6 x 480GB SSDs)” specified in

the “Hardware Requirements” is an empirical value also based on  5 servers?

Thanks again,

Frank

 

From: CARVER, PAUL [mailto:pc2929@...]
Sent: Thursday, June 4, 2020 10:02 PM
To:
蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@...>; technical-discuss@...
Cc:
詹子儀 Andy Chan <andychan@...>; Sean Xie <sean.xie@...>
Subject: RE: About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

 

Here are my answers. Nokia provided the original seed code for REC/TA based on previous closed source work they had done, so there might also be historical factors that I’m not aware of.

 

1)      Yes, there are several components in the system that use a quorum based mechanism where you need more than 50% of the total components to be in communications with each other in order to differentiate between a failure vs a split brain scenario. That effectively makes the HA minimum 3 instead of 2. There could conceivably be any odd number greater than 1 for HA but all of our testing has been based on 3.

 

2)      Yes, “node” is referring to physical servers. Our focus on REC/TA is on integration testing a full platform suitable for production deployment in edge locations where there isn’t a pre-existing general purpose cloud. The distinction between RIC (an O-RAN SC project) and REC (an Akraino project) is that RIC is purely software whereas REC is concerned with providing the cloud infrastructure hardware on top of which the RIC can be deployed.

 

3)      For the joint work between Nokia and AT&T to create the initial release of REC/TA we selected the Nokia OpenEdge chassis based platform as the reference hardware platform for our testing. We also did some testing on Dell and HP hardware and have subsequently seen the support for Ampere’s ARM servers added, but we started with the Nokia OpenEdge which packages 5 servers into a 3U chassis. As such, the vast majority of our testing has been on clusters of 5 servers where 3 of them are controllers and the other 2 servers are not controllers. Our use case is primarily on deploying a large number of small clusters in many different edge locations rather than general purpose datacenter cloud, so the deployment unit of 5 servers in a 3U chassis has been a convenient size.

 

4)      I don’t think we’re currently using DPDK. REC/TA had its roots in a previous Nokia closed source OpenStack deployment system, so Open vSwitch DPDK support was provided with OpenStack Neutron, but REC/TA is not intended as an OpenStack system. It is intended as a pure Kubernetes system that makes use of a few OpenStack components such as Ironic for baremetal deployment and Keystone for authentication.

 

5)      The kernel version was selected specifically after some benchmarking of latency. It’s not a realtime kernel, but latency is a major consideration for the RIC which is the first and primary application that we designed REC to support. That’s not to say that it’s the only kernel version that will support the latency requirements, but if you use a different kernel then it would definitely be a good idea to run some latency benchmarks. I would have to dig through some old documentation to refresh my memory on what tests we ran. I know cyclictest is a popular latency benchmark but I’m pretty sure it wasn’t the only one that we compared.

 

From: technical-discuss@... <technical-discuss@...> On Behalf Of ??? FRANK C. D. TSAI, Ph.D.
Sent: Thursday, June 4, 2020 06:58
To: technical-discuss@...
Cc:
詹子儀 Andy Chan <andychan@...>; Sean Xie <sean.xie@...>
Subject: [Akraino Technical-Discuss] About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

 

Dear community,

 

 

I was wondering if I can sort out some of my questions of Akraino (while I’m using Adlink Adlink H/W ALPS-7400 as the H/W platform for REC, which is a 4U network appliance with 4 computing nodes).

 

Per REC Installation Guide (*), controller nodes (3 required), worker nodes (all optional)

 

1.      Is the purpose of 3 controller nodes for HA (high availability)? 

[note: the reason I suppose so is because our past experience with StarlingX using two controller nodes for HA]

 

2.      When we use the term “node” in the document, do we mean physical nodes? Or it can be logical nodes (like VM)?

[note; I suppose it’s physical nodes because in the same page Hardware Requirements, it reads “Minimum of 3 nodes” and if it’s for HA, we need physical node]

 

3.      When we say “controller nodes (3 required), worker nodes (all optional)”, do we mean to let the controller nodes play dual role also as worker nodes by running applications on top of any controller node? 

 

Or we mean to say that REC deployment must have exactly 3 controller nodes, and after the 3 controller nodes are established, we can (and we should) then gradually add worker nodes to scale out?

 

4.      Is it a must for the NIC of a worker node to support DPDK? if no DPDK is supported, what would be the most significant impact (like some function XYZ is then not available) ?

 

5.      Is using 4.14 is simply because 4.14 newer than 3.10?  Or, it’s because some components of REC must rely on 4.14?

 

[note: I so inquire is because some ALPS-7400 peripheral driver is currently compatible with kernel 3.10, yet Build-237 uses kernel 4.14.  After downloading the ISO (Build-237), we plan to replace the kernel modules from 4.14 to 3.10.]

 

Thank you very much for your guidance,

 

 

Sincerely,

Frank

 

(*) https://wiki.akraino.org/display/AK/REC+Installation+Guide#RECInstallationGuide-HardwareRequirements

 

 


Re: About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

Paul Carver
 

Here are my answers. Nokia provided the original seed code for REC/TA based on previous closed source work they had done, so there might also be historical factors that I’m not aware of.

 

  1. Yes, there are several components in the system that use a quorum based mechanism where you need more than 50% of the total components to be in communications with each other in order to differentiate between a failure vs a split brain scenario. That effectively makes the HA minimum 3 instead of 2. There could conceivably be any odd number greater than 1 for HA but all of our testing has been based on 3.

 

  1. Yes, “node” is referring to physical servers. Our focus on REC/TA is on integration testing a full platform suitable for production deployment in edge locations where there isn’t a pre-existing general purpose cloud. The distinction between RIC (an O-RAN SC project) and REC (an Akraino project) is that RIC is purely software whereas REC is concerned with providing the cloud infrastructure hardware on top of which the RIC can be deployed.

 

  1. For the joint work between Nokia and AT&T to create the initial release of REC/TA we selected the Nokia OpenEdge chassis based platform as the reference hardware platform for our testing. We also did some testing on Dell and HP hardware and have subsequently seen the support for Ampere’s ARM servers added, but we started with the Nokia OpenEdge which packages 5 servers into a 3U chassis. As such, the vast majority of our testing has been on clusters of 5 servers where 3 of them are controllers and the other 2 servers are not controllers. Our use case is primarily on deploying a large number of small clusters in many different edge locations rather than general purpose datacenter cloud, so the deployment unit of 5 servers in a 3U chassis has been a convenient size.

 

  1. I don’t think we’re currently using DPDK. REC/TA had its roots in a previous Nokia closed source OpenStack deployment system, so Open vSwitch DPDK support was provided with OpenStack Neutron, but REC/TA is not intended as an OpenStack system. It is intended as a pure Kubernetes system that makes use of a few OpenStack components such as Ironic for baremetal deployment and Keystone for authentication.

 

  1. The kernel version was selected specifically after some benchmarking of latency. It’s not a realtime kernel, but latency is a major consideration for the RIC which is the first and primary application that we designed REC to support. That’s not to say that it’s the only kernel version that will support the latency requirements, but if you use a different kernel then it would definitely be a good idea to run some latency benchmarks. I would have to dig through some old documentation to refresh my memory on what tests we ran. I know cyclictest is a popular latency benchmark but I’m pretty sure it wasn’t the only one that we compared.

 

From: technical-discuss@... <technical-discuss@...> On Behalf Of ??? FRANK C. D. TSAI, Ph.D.
Sent: Thursday, June 4, 2020 06:58
To: technical-discuss@...
Cc:
詹子儀 Andy Chan <andychan@...>; Sean Xie <sean.xie@...>
Subject: [Akraino Technical-Discuss] About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

 

Dear community,

 

 

I was wondering if I can sort out some of my questions of Akraino (while I’m using Adlink Adlink H/W ALPS-7400 as the H/W platform for REC, which is a 4U network appliance with 4 computing nodes).

 

Per REC Installation Guide (*), controller nodes (3 required), worker nodes (all optional)

 

  1. Is the purpose of 3 controller nodes for HA (high availability)? 

[note: the reason I suppose so is because our past experience with StarlingX using two controller nodes for HA]

 

  1. When we use the term “node” in the document, do we mean physical nodes? Or it can be logical nodes (like VM)?

[note; I suppose it’s physical nodes because in the same page Hardware Requirements, it reads “Minimum of 3 nodes” and if it’s for HA, we need physical node]

 

  1. When we say “controller nodes (3 required), worker nodes (all optional)”, do we mean to let the controller nodes play dual role also as worker nodes by running applications on top of any controller node? 

 

Or we mean to say that REC deployment must have exactly 3 controller nodes, and after the 3 controller nodes are established, we can (and we should) then gradually add worker nodes to scale out?

 

  1. Is it a must for the NIC of a worker node to support DPDK? if no DPDK is supported, what would be the most significant impact (like some function XYZ is then not available) ?

 

  1. Is using 4.14 is simply because 4.14 newer than 3.10?  Or, it’s because some components of REC must rely on 4.14?

 

[note: I so inquire is because some ALPS-7400 peripheral driver is currently compatible with kernel 3.10, yet Build-237 uses kernel 4.14.  After downloading the ISO (Build-237), we plan to replace the kernel modules from 4.14 to 3.10.]

 

Thank you very much for your guidance,

 

 

Sincerely,

Frank

 

(*) https://wiki.akraino.org/display/AK/REC+Installation+Guide#RECInstallationGuide-HardwareRequirements

 

 


About Akraino REC intallation (min 3 controller nodes with 192GB minimum per node)

蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@...>
 

Dear community,

 

 

I was wondering if I can sort out some of my questions of Akraino (while I’m using Adlink Adlink H/W ALPS-7400 as the H/W platform for REC, which is a 4U network appliance with 4 computing nodes).

 

Per REC Installation Guide (*), controller nodes (3 required), worker nodes (all optional)

 

1.      Is the purpose of 3 controller nodes for HA (high availability)? 

[note: the reason I suppose so is because our past experience with StarlingX using two controller nodes for HA]

 

2.      When we use the term “node” in the document, do we mean physical nodes? Or it can be logical nodes (like VM)?

[note; I suppose it’s physical nodes because in the same page Hardware Requirements, it reads “Minimum of 3 nodes” and if it’s for HA, we need physical node]

 

3.      When we say “controller nodes (3 required), worker nodes (all optional)”, do we mean to let the controller nodes play dual role also as worker nodes by running applications on top of any controller node? 

 

Or we mean to say that REC deployment must have exactly 3 controller nodes, and after the 3 controller nodes are established, we can (and we should) then gradually add worker nodes to scale out?

 

4.      Is it a must for the NIC of a worker node to support DPDK? if no DPDK is supported, what would be the most significant impact (like some function XYZ is then not available) ?

 

5.      Is using 4.14 is simply because 4.14 newer than 3.10?  Or, it’s because some components of REC must rely on 4.14?

 

[note: I so inquire is because some ALPS-7400 peripheral driver is currently compatible with kernel 3.10, yet Build-237 uses kernel 4.14.  After downloading the ISO (Build-237), we plan to replace the kernel modules from 4.14 to 3.10.]

 

Thank you very much for your guidance,

 

 

Sincerely,

Frank

 

(*) https://wiki.akraino.org/display/AK/REC+Installation+Guide#RECInstallationGuide-HardwareRequirements

 

 


Upcoming Event: Akraino Technical Community Call (Weekly) - Thu, 06/04/2020 1:00pm-2:00pm #cal-reminder

technical-discuss@lists.akraino.org Calendar <technical-discuss@...>
 

Reminder: Akraino Technical Community Call (Weekly)

When: Thursday, 4 June 2020, 1:00pm to 2:00pm, (GMT+00:00) UTC

Where:https://zoom.us/j/919148693

View Event

Organizer: technical-discuss@...

Description:

Akraino Technical Community Call: TSC updates to technical community and deeper dives into topics as applicable. Meeting content posted to Technical Community Wiki.
Meeting Lead: Kandan Kathirvel, Akraino TSC Chair


Akraino Edge Stack is inviting you to a scheduled Zoom meeting.
Join from PC, Mac, Linux, iOS or Android:
https://zoom.us/j/919148693
Or iPhone one-tap : US: +16699006833,,919148693# or +16465588656,,919148693#
Or Telephone: Dial(for higher quality, dial a number based on your current location):
US: +1 669 900 6833 or +1 646 558 8656 or +1 877 369 0926 (Toll Free) or +1 855 880 1246 (Toll Free)
Meeting ID: 919 148 693
International numbers available:
https://zoom.us/u/adnlim1pfM


Upcoming Event: Akraino Technical Community Call (Weekly) - Thu, 06/04/2020 1:00pm-2:00pm #cal-reminder

technical-discuss@lists.akraino.org Calendar <technical-discuss@...>
 

Reminder: Akraino Technical Community Call (Weekly)

When: Thursday, 4 June 2020, 1:00pm to 2:00pm, (GMT+00:00) UTC

Where:https://zoom.us/j/919148693

View Event

Organizer: technical-discuss@...

Description:

Akraino Technical Community Call: TSC updates to technical community and deeper dives into topics as applicable. Meeting content posted to Technical Community Wiki.
Meeting Lead: Kandan Kathirvel, Akraino TSC Chair


Akraino Edge Stack is inviting you to a scheduled Zoom meeting.
Join from PC, Mac, Linux, iOS or Android:
https://zoom.us/j/919148693
Or iPhone one-tap : US: +16699006833,,919148693# or +16465588656,,919148693#
Or Telephone: Dial(for higher quality, dial a number based on your current location):
US: +1 669 900 6833 or +1 646 558 8656 or +1 877 369 0926 (Toll Free) or +1 855 880 1246 (Toll Free)
Meeting ID: 919 148 693
International numbers available:
https://zoom.us/u/adnlim1pfM


Re: [REC]Question about Hardware Requiement

Tina Tsou
 

Dear Andy et al,

You may want to join REC weekly meeting to discuss.

Radio Edge Cloud and SEBA weekly meeting

When:
Thursday, 28 May 2020
1:00pm to 2:00pm
(GMT+00:00) UTC

Where:
https://zoom.us/j/941470503


Thank you,
Tina ^ ^

On May 28, 2020, at 6:32 AM, 詹子儀 Andy Chan via lists.akraino.org <andychan=iii.org.tw@...> wrote:



Dear Community

 

Hi there,

 

Here I planning to build REC. Before the installation, I am wondering that each server spec. In the install guide(https://wiki.akraino.org/display/AK/REC+Installation+Guide#RECInstallationGuide-HardwareRequirements:), the Hardware Requirements say that.

 

l   Minimum of 3 nodes

l   Total Physical Compute Cores:  60 (120 vCPUs)

l   Total Physical Compute Memory:  192GB minimum per node

l   Total SSD-based OS Storage:  2.8 TB (6 x 480GB SSDs)

l   Total Application-based Raw Storage:  5.7 TB (6 x 960GB SSD0

l   Networking Per Server:     Apps - 2 x 25GbE (per Server) and DCIM - 2 x 10GbE + 1 1Gbt (shared)

 

So my question is that what is “Total” Means? Is that mean three nodes?

 

Another question is that what is the difference between SSD-based OS Storage and Application-based Raw Storage, what data will store in SSD-based OS Storage? What data will store in Application-based Raw Storage.

 

Best Regards,

Andy

........................................................
詹子儀 Andy Chan

Email: andychan@...

 

IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.


[REC]Question about Hardware Requiement

詹子儀 Andy Chan <andychan@...>
 

Dear Community

 

Hi there,

 

Here I planning to build REC. Before the installation, I am wondering that each server spec. In the install guide(https://wiki.akraino.org/display/AK/REC+Installation+Guide#RECInstallationGuide-HardwareRequirements:), the Hardware Requirements say that.

 

l   Minimum of 3 nodes

l   Total Physical Compute Cores:  60 (120 vCPUs)

l   Total Physical Compute Memory:  192GB minimum per node

l   Total SSD-based OS Storage:  2.8 TB (6 x 480GB SSDs)

l   Total Application-based Raw Storage:  5.7 TB (6 x 960GB SSD0

l   Networking Per Server:     Apps - 2 x 25GbE (per Server) and DCIM - 2 x 10GbE + 1 1Gbt (shared)

 

So my question is that what is “Total” Means? Is that mean three nodes?

 

Another question is that what is the difference between SSD-based OS Storage and Application-based Raw Storage, what data will store in SSD-based OS Storage? What data will store in Application-based Raw Storage.

 

Best Regards,

Andy

........................................................
詹子儀 Andy Chan

Email: andychan@...

 


Upcoming Event: Akraino Technical Community Call (Weekly) - Thu, 05/28/2020 1:00pm-2:00pm #cal-reminder

technical-discuss@lists.akraino.org Calendar <technical-discuss@...>
 

Reminder: Akraino Technical Community Call (Weekly)

When: Thursday, 28 May 2020, 1:00pm to 2:00pm, (GMT+00:00) UTC

Where:https://zoom.us/j/919148693

View Event

Organizer: technical-discuss@...

Description:

Akraino Technical Community Call: TSC updates to technical community and deeper dives into topics as applicable. Meeting content posted to Technical Community Wiki.
Meeting Lead: Kandan Kathirvel, Akraino TSC Chair


Akraino Edge Stack is inviting you to a scheduled Zoom meeting.
Join from PC, Mac, Linux, iOS or Android:
https://zoom.us/j/919148693
Or iPhone one-tap : US: +16699006833,,919148693# or +16465588656,,919148693#
Or Telephone: Dial(for higher quality, dial a number based on your current location):
US: +1 669 900 6833 or +1 646 558 8656 or +1 877 369 0926 (Toll Free) or +1 855 880 1246 (Toll Free)
Meeting ID: 919 148 693
International numbers available:
https://zoom.us/u/adnlim1pfM


Upcoming Event: Akraino Technical Community Call (Weekly) - Thu, 05/28/2020 1:00pm-2:00pm #cal-reminder

technical-discuss@lists.akraino.org Calendar <technical-discuss@...>
 

Reminder: Akraino Technical Community Call (Weekly)

When: Thursday, 28 May 2020, 1:00pm to 2:00pm, (GMT+00:00) UTC

Where:https://zoom.us/j/919148693

View Event

Organizer: technical-discuss@...

Description:

Akraino Technical Community Call: TSC updates to technical community and deeper dives into topics as applicable. Meeting content posted to Technical Community Wiki.
Meeting Lead: Kandan Kathirvel, Akraino TSC Chair


Akraino Edge Stack is inviting you to a scheduled Zoom meeting.
Join from PC, Mac, Linux, iOS or Android:
https://zoom.us/j/919148693
Or iPhone one-tap : US: +16699006833,,919148693# or +16465588656,,919148693#
Or Telephone: Dial(for higher quality, dial a number based on your current location):
US: +1 669 900 6833 or +1 646 558 8656 or +1 877 369 0926 (Toll Free) or +1 855 880 1246 (Toll Free)
Meeting ID: 919 148 693
International numbers available:
https://zoom.us/u/adnlim1pfM


Upcoming Event: Akraino Technical Community Call (Weekly) - Thu, 05/21/2020 1:00pm-2:00pm #cal-reminder

technical-discuss@lists.akraino.org Calendar <technical-discuss@...>
 

Reminder: Akraino Technical Community Call (Weekly)

When: Thursday, 21 May 2020, 1:00pm to 2:00pm, (GMT+00:00) UTC

Where:https://zoom.us/j/919148693

View Event

Organizer: technical-discuss@...

Description:

Akraino Technical Community Call: TSC updates to technical community and deeper dives into topics as applicable. Meeting content posted to Technical Community Wiki.
Meeting Lead: Kandan Kathirvel, Akraino TSC Chair


Akraino Edge Stack is inviting you to a scheduled Zoom meeting.
Join from PC, Mac, Linux, iOS or Android:
https://zoom.us/j/919148693
Or iPhone one-tap : US: +16699006833,,919148693# or +16465588656,,919148693#
Or Telephone: Dial(for higher quality, dial a number based on your current location):
US: +1 669 900 6833 or +1 646 558 8656 or +1 877 369 0926 (Toll Free) or +1 855 880 1246 (Toll Free)
Meeting ID: 919 148 693
International numbers available:
https://zoom.us/u/adnlim1pfM


Upcoming Event: Akraino Technical Community Call (Weekly) - Thu, 05/21/2020 1:00pm-2:00pm #cal-reminder

technical-discuss@lists.akraino.org Calendar <technical-discuss@...>
 

Reminder: Akraino Technical Community Call (Weekly)

When: Thursday, 21 May 2020, 1:00pm to 2:00pm, (GMT+00:00) UTC

Where:https://zoom.us/j/919148693

View Event

Organizer: technical-discuss@...

Description:

Akraino Technical Community Call: TSC updates to technical community and deeper dives into topics as applicable. Meeting content posted to Technical Community Wiki.
Meeting Lead: Kandan Kathirvel, Akraino TSC Chair


Akraino Edge Stack is inviting you to a scheduled Zoom meeting.
Join from PC, Mac, Linux, iOS or Android:
https://zoom.us/j/919148693
Or iPhone one-tap : US: +16699006833,,919148693# or +16465588656,,919148693#
Or Telephone: Dial(for higher quality, dial a number based on your current location):
US: +1 669 900 6833 or +1 646 558 8656 or +1 877 369 0926 (Toll Free) or +1 855 880 1246 (Toll Free)
Meeting ID: 919 148 693
International numbers available:
https://zoom.us/u/adnlim1pfM


Re: [REC]Install and Test Problem.

詹子儀 Andy Chan <andychan@...>
 

Dear Tapio and Wenhui,

 

Thank about that information you gave me,

 

but still, there has a problem, I use Qemu to install the VM version for REC, somehow it processes to first step and fails.

below is the error message in the VM.

 

And According to the mail, it seems that you have already tested the VM personal.

In the figure below, that is my setting parameters, if possible can I asked you that what the VM parameters you set to pass the REC installed.

 

==

Host OS: Ubuntu 16.04

VM CPUs: 2

VM RAM: 4GB

VM Memory: 20

NIC: Virto

Best Regards,

Andy

From: Wenhui Zhang <wenhui@...>
Sent: Friday, May 15, 2020 10:54 AM
To: Tallgren, Tapio (Nokia - FI/Espoo) <tapio.tallgren@...>
Cc:
詹子儀 Andy Chan <andychan@...>; technical-discuss@...; 蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@...>
Subject: Re: [Akraino Technical-Discuss] [REC]Install and Test Problem.

 

Thanks Tapio!

It works !

Wenhui 

 

On Thu, May 14, 2020, 3:48 AM Tallgren, Tapio (Nokia - FI/Espoo) <tapio.tallgren@...> wrote:

FWIW, dpdk should run just fine in a Qemu VM with the virtio poll-mode driver: https://doc.dpdk.org/guides/nics/virtio.html

May require some trickery, though.

-Tapio

________________________________________
From: technical-discuss@... <technical-discuss@...> on behalf of Wenhui via lists.akraino.org <wenhui=gwmail.gwu.edu@...>
Sent: Thursday, May 14, 2020 9:49
To:
詹子儀 Andy Chan
Cc: technical-discuss@...;
蔡其達 FRANK C. D. TSAI, Ph.D.
Subject: Re: [Akraino Technical-Discuss] [REC]Install and Test Problem.

Sorry that for the former mistake I made.
Thanks Jim for correcting the error.

If you have a physical NIC card , then you could go ahead and test it on your computer without VM.

If you do not have the physical NIC card, which supports DPDK, then it might be hard to emulate the hardware in qemu.

Please let me know if you happens to know some qemu which could do DPDK based NIC emulation.

Thanks,
Wenhui

On Thu, May 14, 2020, 1:26 AM Wenhui Zhang <wenhui@...<mailto:wenhui@...>> wrote:

I am not aware of any virtual machine (qemu or virtualbox) , which supports dpdk right now.

Thus, there might be a problem with testing through VMs.


To understand the code, below folders are important, it is built with golang, any laptop or server should be fine.

com.golog    it.dep   nonrtric  portal.ric-dashboard  ric-app.mc<http://ric-app.mc>      ric-plt.dbaas          ric-plt.lib.rmr                  ric-plt.sdl                  ric-plt.tracelibcpp  scp.ocu.openlte
com.gs-lite  it.dev<http://it.dev>   oam       pti.rtp               ric-app.ml<http://ric-app.ml>      ric-plt.e2             ric-plt.nodeb-rnib               ric-plt.sdlgo                ric-plt.tracelibgo   sim.o1-interface
com.log      it.otf   o-du.l2   ric-app.admin         ric-plt.a1      ric-plt.e2mgr          ric-plt.resource-status-manager  ric-plt.streaming-protobufs  ric-plt.vespamgr
com.pylog    it.test  o-du.phy  ric-app.kpimon        ric-plt.appmgr  ric-plt.jaegeradapter  ric-plt.rtmgr                    ric-plt.submgr               ric-plt.xapp-frame


Please ref to here for hardware : https://docs.o-ran-sc.org/projects/o-ran-sc-pti-rtp/en/latest/installation-guide.html#abstract

Following minimum hardware requirements must be met for installation of O-RAN INF image:

HW Aspect       Requirement
# of servers    1
CPU     2
RAM     4G
Disk    20G
NICs    1




On Thu, May 14, 2020 at 12:27 AM
詹子儀 Andy Chan <andychan@...<mailto:andychan@...>> wrote:
Hi Team,

Right now, I'm trying to install REC in virtual machine (using VirtualBox),but it seem that HDD() has problem, so the install can't process. My question is that possible to  install in virtual machine to test ? and also how can I to set parameter of virtual machine ?

By the way, I was wonder that in hardware type there was a python file the is hw_unhown.py, it seem that REC is already defend the un-test server. So that we don't need to build the install.iso, and my question is that is it necessary to build the install.iso file if we using the different hardware ? If it is necessary,  there is a lot of git project in server's space, what is the key projects i need to download ?

BR,
Andy
........................................................
詹子儀 | Andy Chan
Office: +886 6607-3242
Email: andychan@...<mailto:andychan@...>


-----Original Message-----
From: CARVER, PAUL <pc2929@...<mailto:pc2929@...>>
Sent: Wednesday, May 6, 2020 8:02 PM
To:
詹子儀 Andy Chan <andychan@...<mailto:andychan@...>>; technical-discuss@...<mailto:technical-discuss@...>
Cc:
蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@...<mailto:ftsai@...>>
Subject: RE: [REC]Install and Test Problem.

If you’re doing a semi-manual install you only need install.iso. If you want to setup a Regional Controller and use it to automate the install process then you need bootcd.iso in addition to install.iso. I just downloaded install.iso successfully and verified the sha256 hash:



~$ wget https://nexus.akraino.org/content/repositories/images-snapshots/TA/release-1/images/237/install.iso

--2020-05-06 07:53:34--  https://nexus.akraino.org/content/repositories/images-snapshots/TA/release-1/images/237/install.iso

Resolving nexus.akraino.org<http://nexus.akraino.org> (nexus.akraino.org<http://nexus.akraino.org>)... 162.253.55.36

Connecting to nexus.akraino.org<http://nexus.akraino.org> (nexus.akraino.org<http://nexus.akraino.org>)|162.253.55.36|:443... connected.

HTTP request sent, awaiting response... 200 OK

Length: 3129999360 (2.9G) [application/x-iso9660-image]

Saving to: ‘install.iso’



install.iso                                                100%[======================================================================================================================================>]   2.92G  18.6MB/s    in 2m 38s



2020-05-06 07:56:12 (18.9 MB/s) - ‘install.iso’ saved [3129999360/3129999360]



~$ sha256sum install.iso

fdc1258ab40a3f54706d96236d2c70645268927d801ac747cf0aa50ee8f16a0a  install.iso

~$



However, if you’re using different hardware then you will almost certainly need to customize the hardware detector. The code is here: https://gerrit.akraino.org/r/gitweb?p=ta/hw-detector.git;a=tree;f=src/hw_detector/hw_types;h=579c0084469052da8ca801ef7a31e29a4aec5a0e;hb=HEAD <https://gerrit.akraino.org/r/gitweb?p=ta/hw-detector.git;a=tree;f=src/hw_detector/hw_types;h=579c0084469052da8ca801ef7a31e29a4aec5a0e;hb=HEAD>



And the instructions for building your own ISO are here: https://wiki.akraino.org/display/AK/How+to+Build+a+REC+or+Telco+Appliance+ISO <https://wiki.akraino.org/display/AK/How+to+Build+a+REC+or+Telco+Appliance+ISO>





From: technical-discuss@...<mailto:technical-discuss@...> <technical-discuss@...<mailto:technical-discuss@...>> On Behalf Of ??? Andy Chan
Sent: Wednesday, May 6, 2020 04:28
To: technical-discuss@...<mailto:technical-discuss@...>
Cc:
蔡其達 FRANK C. D. TSAI, Ph.D. <ftsai@...<mailto:ftsai@...>>
Subject: [Akraino Technical-Discuss] [REC]Install and Test Problem.



Dear Community,



Hello there,



We expect to install REC on the server for testing (Model: ALPS-7400, Vendor: Adlink, CPU: Intel Xeon CPU). At present, when downloading the installation files, I found that there are two image files bootcd.iso and install.iso(https://nexus.akraino.org/content/repositories/images-snapshots/TA/release-1/images/latest/ <https://urldefense.proofpoint.com/v2/url?u=https-3A__nexus.akraino.org_content_repositories_images-2Dsnapshots_TA_release-2D1_images_latest_&d=DwMFog&c=LFYZ-o9_HUMeMTSQicvjIg&r=HBNonG828PGilNRNwXAtdg&m=wVrpGXZLlnttAK54oVhO07GPwv-Gw9SedhQkdwk34hs&s=WAIPqx-qsXgE4iovQ58lS3VgxRcsYUxiBRb6LspffVQ&e=> ), which file should I use?



In addition, in the Nexus download space, we found that the install.iso file could not be downloaded or the download will be terminated when it reaches a certain level.We are a wonder that the file is exist now?



Another question is about Build 9 and Build 237,According to the following instructions(https://wiki.akraino.org/display/AK/REC+Installation+Guide#RECInstallationGuide-Pre-InstallationRequirementsforRECCluster <https://urldefense.proofpoint.com/v2/url?u=https-3A__wiki.akraino.org_display_AK_REC-2BInstallation-2BGuide-23RECInstallationGuide-2DPre-2DInstallationRequirementsforRECCluster&d=DwMFog&c=LFYZ-o9_HUMeMTSQicvjIg&r=HBNonG828PGilNRNwXAtdg&m=wVrpGXZLlnttAK54oVhO07GPwv-Gw9SedhQkdwk34hs&s=34JcJKxziVAPTJ1OciM97_3YkqzldviSrNxhL9wd0dk&e=>  ), we can know to Build 237 that REC has improved on Dell Server and Open Edge. What if we are not using the server that recommended by the website, can we install with REC?



Sincerely,

Andy








--
V/R,

Wenhui Zhang

Email: wenhui@...<mailto:wenhui@...>
           Telephone: 1-(919) 945 9921