z.archive.alto-dev@lists.opendaylight.org

Y. Richard Yang <yang.r.yang@...>

Austin,

Can you please let Azher know the issues that you fix in l2switch?

Thanks!

Richard

On Friday, October 9, 2015, Azher Mughal <azher@...> wrote:

Hi Richard,

On 10/9/2015 1:36 PM, Y. Richard Yang wrote:

Azher,

Problems with real switches can always be hard to debug. We do not have this mentioned switch, and hence it is great to know about this issue. I feel that such issues can be addressed by a more robust l2switch module, which we can propose to l2switch, make sense?

I agree. This specific issue seems to be related with just the Dell Z9000 only. This has the OpenFlow implementation but code is only under maintenance. So hopefully Dell will release some upgraded firmware to get this fixed.

I remember you mentioned some of the code improvements in l2switch, can you please list the bug fixes or improvements you have made ? We are getting ready for the SC15 conference and the current arp manager of the ODL has issues, and it would be nice to have a better error free l2switch from you. Please let me know.

Thanks
-Azher

Richard

On Fri, Oct 9, 2015 at 11:03 AM, Azher Mughal <azher@...> wrote:

Sorry for not replying earlier, I was trying to understand more on what was happening with the actual switches.

One of the switch in the topology is Dell Z9000, this switch has buggy OF implementation and was passing the hosts and link info from one instance to other, thus completely confusing the whole topology.

I might have more questions regarding the Python flow insertion into Lithium :)

Thanks
-Azher

On 9/30/2015 11:41 PM, Junzhuo Wang wrote:

Hi Azher,

As I know the latest version of L2Switch is 0.3.0-SNAPSHOT. So may I ask your the source of the code? Gerrit or Github? And could you give me the real topology of your network? I’ll test it because I’m not sure the L2Switch work fine with multi-NIC hosts.

If you want to push your own flow entry. The setting should be fine but I saw no default flow in your switches. Because normally all switch should have a default flow entry. The HostTracker may not working if the LLDP or ARP packets are not sent to controller properly.

Regards,

Junzhuo

On Sep 30, 2015, at 10:25 PM, Azher Mughal <azher@...> wrote:

Hi Junzhuo,

Please see this topology. I am using the daily snapshots of Lithium 0.3.2
MailScanner warning: numerical links are often malicious: http://131.215.207.90:8080/index.html#/topology

The problem I see is when l2-switch (and host tracker) is enabled , hosts start to appear on different nodes as you can see in the picture below. Our intention to use l2-switch is to get the topology and hosts automatically discovered, but not to let it push the flow rules by itself. Thus I apply these two configs to ODL:

is-proactive-flood-mode=false
is-learning-only-mode=true

But still hosts appear to multiple nodes. Any suggestions ?

Thanks
-AZher
<cgfagbhi.png>

On 9/30/2015 6:03 AM, Junzhuo Wang wrote:

Hi Azher,

I could help you to solve the L2Switch issue but I need to know which branch you’re working and your test condition. The problem you mentioned seems not appears here. But we only test the l2switch on mininet so I’m not sure if the HostTracker really works with a real OpenFlow switch and I modified the L2Switch based on stable/lithium tag.

If you have any question or information. Feel free to mail me :-).

Regards,

Junzhuo

On Sep 30, 2015, at 8:34 PM, Y. Richard Yang <yang.r.yang@...> wrote:

Hi Azher,

It was great to you yesterday.

I cc'd the alto-team.

Jensen/Junzhuo/Tony: Azher from CalTech mentioned that an issue they are facing is that ODL l2switch (HostTracker) made one host appears at multiple locations. Our version does not have this problem, I believe. Could you please engage Azher to solve this problem?

Thanks a lot!

Richard

On Wednesday, September 30, 2015, Azher Mughal <azher@...> wrote:

Hi Richard,

Nice talking with you yesterday. Do yo mind sharing the project with us, so I can download and try it out. Current issues as I mentioned are hosts seen connected to various nodes which creates a total confusion.

Thanks
-Azher

--
Richard

--
You received this message because you are subscribed to the Google Groups "alto-design" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alto-design+unsubscribe@....
To post to this group, send email to alto-design@....
To view this discussion on the web visit https://groups.google.com/d/msgid/alto-design/CANUuoLo%2B30hrds%2BNWcJDO3yYtb7uN23yeEsPN6L4aUk%3DOuS55A%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
alto-dev mailing list
alto-dev@...
https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.opendaylight.org_mailman_listinfo_alto-2Ddev&d=AwICAg&c=-dg2m7zWuuDZ0MUcV7Sdqw&r=4G36iiEVb2m_v-0RnP2gx9KZJjYQgfvrOCE3789JGIA&m=HHh44QXXCExMFYWl9lOY-9fNvoHW4RwaTYEwEN1HFpw&s=p4_IIiVWthIFUwZ-Vcg5VgqQEiYlRWCbHg0ypY459uA&e=

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

--
Richard

Azher Mughal <azher@...>

#256

Attached is the topology we are building for SC15.

Thanks
-Azher

On 10/9/2015 6:26 PM, Y. Richard Yang wrote:

Show quoted text

Austin,

Can you please let Azher know the issues that you fix in l2switch?

Thanks!

Richard

On Friday, October 9, 2015, Azher Mughal <azher@...> wrote:

Hi Richard,

On 10/9/2015 1:36 PM, Y. Richard Yang wrote:

Azher,

Problems with real switches can always be hard to debug. We do not have this mentioned switch, and hence it is great to know about this issue. I feel that such issues can be addressed by a more robust l2switch module, which we can propose to l2switch, make sense?

I agree. This specific issue seems to be related with just the Dell Z9000 only. This has the OpenFlow implementation but code is only under maintenance. So hopefully Dell will release some upgraded firmware to get this fixed.

I remember you mentioned some of the code improvements in l2switch, can you please list the bug fixes or improvements you have made ? We are getting ready for the SC15 conference and the current arp manager of the ODL has issues, and it would be nice to have a better error free l2switch from you. Please let me know.

Thanks
-Azher

Richard

On Fri, Oct 9, 2015 at 11:03 AM, Azher Mughal <azher@...> wrote:

Sorry for not replying earlier, I was trying to understand more on what was happening with the actual switches.

One of the switch in the topology is Dell Z9000, this switch has buggy OF implementation and was passing the hosts and link info from one instance to other, thus completely confusing the whole topology.

I might have more questions regarding the Python flow insertion into Lithium :)

Thanks
-Azher

On 9/30/2015 11:41 PM, Junzhuo Wang wrote:

Hi Azher,

As I know the latest version of L2Switch is 0.3.0-SNAPSHOT. So may I ask your the source of the code? Gerrit or Github? And could you give me the real topology of your network? I’ll test it because I’m not sure the L2Switch work fine with multi-NIC hosts.

If you want to push your own flow entry. The setting should be fine but I saw no default flow in your switches. Because normally all switch should have a default flow entry. The HostTracker may not working if the LLDP or ARP packets are not sent to controller properly.

Regards,

Junzhuo

On Sep 30, 2015, at 10:25 PM, Azher Mughal <azher@...> wrote:

Hi Junzhuo,

Please see this topology. I am using the daily snapshots of Lithium 0.3.2
MailScanner warning: numerical links are often malicious: http://131.215.207.90:8080/index.html#/topology

The problem I see is when l2-switch (and host tracker) is enabled , hosts start to appear on different nodes as you can see in the picture below. Our intention to use l2-switch is to get the topology and hosts automatically discovered, but not to let it push the flow rules by itself. Thus I apply these two configs to ODL:

is-proactive-flood-mode=false
is-learning-only-mode=true

But still hosts appear to multiple nodes. Any suggestions ?

Thanks
-AZher
<cgfagbhi.png>

On 9/30/2015 6:03 AM, Junzhuo Wang wrote:

Hi Azher,

I could help you to solve the L2Switch issue but I need to know which branch you’re working and your test condition. The problem you mentioned seems not appears here. But we only test the l2switch on mininet so I’m not sure if the HostTracker really works with a real OpenFlow switch and I modified the L2Switch based on stable/lithium tag.

If you have any question or information. Feel free to mail me :-).

Regards,

Junzhuo

On Sep 30, 2015, at 8:34 PM, Y. Richard Yang <yang.r.yang@...> wrote:

Hi Azher,

It was great to you yesterday.

I cc'd the alto-team.

Jensen/Junzhuo/Tony: Azher from CalTech mentioned that an issue they are facing is that ODL l2switch (HostTracker) made one host appears at multiple locations. Our version does not have this problem, I believe. Could you please engage Azher to solve this problem?

Thanks a lot!

Richard

On Wednesday, September 30, 2015, Azher Mughal <azher@...> wrote:

Hi Richard,

Nice talking with you yesterday. Do yo mind sharing the project with us, so I can download and try it out. Current issues as I mentioned are hosts seen connected to various nodes which creates a total confusion.

Thanks
-Azher

--
Richard

--
You received this message because you are subscribed to the Google Groups "alto-design" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alto-design+unsubscribe@....
To post to this group, send email to alto-design@....
To view this discussion on the web visit MailScanner has detected a possible fraud attempt from "urldefense.proofpoint.com" claiming to be https://groups.google.com/d/msgid/alto-design/CANUuoLo%2B30hrds%2BNWcJDO3yYtb7uN23yeEsPN6L4aUk%3DOuS55A%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
alto-dev mailing list
alto-dev@...
https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.opendaylight.org_mailman_listinfo_alto-2Ddev&d=AwICAg&c=-dg2m7zWuuDZ0MUcV7Sdqw&r=4G36iiEVb2m_v-0RnP2gx9KZJjYQgfvrOCE3789JGIA&m=HHh44QXXCExMFYWl9lOY-9fNvoHW4RwaTYEwEN1HFpw&s=p4_IIiVWthIFUwZ-Vcg5VgqQEiYlRWCbHg0ypY459uA&e=

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

--
Richard

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

Austin

#257

Hi Azher,

Sorry for the delay. I modified the L2Switch not for fixing bugs, but for improving it. The original L2Switch uses spinning tree to flood every packet on the network, which is not effective. And ODL does not have a service to provide routing information. So basically what I did is:

1. Replace spin tree with shortest path algorithm to build a p2p flow between two hosts;

2. Add APIs to provide routing information.

An issue is that openflow-plugin, a module what L2Switch depends on, has some bugs. So L2Switch may not compact with every switch. I'll try to solve your problem by reading the spec of Dell Z9100 and the docs of openflow-plugin.

Regards,

Austin (Junzhuo)

2015-10-10 9:26 GMT+08:00 Y. Richard Yang <yang.r.yang@...>:

Show quoted text

Austin,

Can you please let Azher know the issues that you fix in l2switch?

Thanks!

Richard

On Friday, October 9, 2015, Azher Mughal <azher@...> wrote:

Hi Richard,

On 10/9/2015 1:36 PM, Y. Richard Yang wrote:

Azher,

Problems with real switches can always be hard to debug. We do not have this mentioned switch, and hence it is great to know about this issue. I feel that such issues can be addressed by a more robust l2switch module, which we can propose to l2switch, make sense?

I agree. This specific issue seems to be related with just the Dell Z9000 only. This has the OpenFlow implementation but code is only under maintenance. So hopefully Dell will release some upgraded firmware to get this fixed.

I remember you mentioned some of the code improvements in l2switch, can you please list the bug fixes or improvements you have made ? We are getting ready for the SC15 conference and the current arp manager of the ODL has issues, and it would be nice to have a better error free l2switch from you. Please let me know.

Thanks
-Azher

Richard

On Fri, Oct 9, 2015 at 11:03 AM, Azher Mughal <azher@...> wrote:

Sorry for not replying earlier, I was trying to understand more on what was happening with the actual switches.

One of the switch in the topology is Dell Z9000, this switch has buggy OF implementation and was passing the hosts and link info from one instance to other, thus completely confusing the whole topology.

I might have more questions regarding the Python flow insertion into Lithium :)

Thanks
-Azher

On 9/30/2015 11:41 PM, Junzhuo Wang wrote:

Hi Azher,

As I know the latest version of L2Switch is 0.3.0-SNAPSHOT. So may I ask your the source of the code? Gerrit or Github? And could you give me the real topology of your network? I’ll test it because I’m not sure the L2Switch work fine with multi-NIC hosts.

If you want to push your own flow entry. The setting should be fine but I saw no default flow in your switches. Because normally all switch should have a default flow entry. The HostTracker may not working if the LLDP or ARP packets are not sent to controller properly.

Regards,

Junzhuo

On Sep 30, 2015, at 10:25 PM, Azher Mughal <azher@...> wrote:

Hi Junzhuo,

Please see this topology. I am using the daily snapshots of Lithium 0.3.2
MailScanner warning: numerical links are often malicious: http://131.215.207.90:8080/index.html#/topology

The problem I see is when l2-switch (and host tracker) is enabled , hosts start to appear on different nodes as you can see in the picture below. Our intention to use l2-switch is to get the topology and hosts automatically discovered, but not to let it push the flow rules by itself. Thus I apply these two configs to ODL:

is-proactive-flood-mode=false
is-learning-only-mode=true

But still hosts appear to multiple nodes. Any suggestions ?

Thanks
-AZher
<cgfagbhi.png>

On 9/30/2015 6:03 AM, Junzhuo Wang wrote:

Hi Azher,

I could help you to solve the L2Switch issue but I need to know which branch you’re working and your test condition. The problem you mentioned seems not appears here. But we only test the l2switch on mininet so I’m not sure if the HostTracker really works with a real OpenFlow switch and I modified the L2Switch based on stable/lithium tag.

If you have any question or information. Feel free to mail me :-).

Regards,

Junzhuo

On Sep 30, 2015, at 8:34 PM, Y. Richard Yang <yang.r.yang@...> wrote:

Hi Azher,

It was great to you yesterday.

I cc'd the alto-team.

Jensen/Junzhuo/Tony: Azher from CalTech mentioned that an issue they are facing is that ODL l2switch (HostTracker) made one host appears at multiple locations. Our version does not have this problem, I believe. Could you please engage Azher to solve this problem?

Thanks a lot!

Richard

On Wednesday, September 30, 2015, Azher Mughal <azher@...> wrote:

Hi Richard,

Nice talking with you yesterday. Do yo mind sharing the project with us, so I can download and try it out. Current issues as I mentioned are hosts seen connected to various nodes which creates a total confusion.

Thanks
-Azher

--
Richard

--
You received this message because you are subscribed to the Google Groups "alto-design" group.
To unsubscribe from this group and stop receiving emails from it, send an email to alto-design+unsubscribe@....
To post to this group, send email to alto-design@....
To view this discussion on the web visit https://groups.google.com/d/msgid/alto-design/CANUuoLo%2B30hrds%2BNWcJDO3yYtb7uN23yeEsPN6L4aUk%3DOuS55A%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
alto-dev mailing list
alto-dev@...
https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.opendaylight.org_mailman_listinfo_alto-2Ddev&d=AwICAg&c=-dg2m7zWuuDZ0MUcV7Sdqw&r=4G36iiEVb2m_v-0RnP2gx9KZJjYQgfvrOCE3789JGIA&m=HHh44QXXCExMFYWl9lOY-9fNvoHW4RwaTYEwEN1HFpw&s=p4_IIiVWthIFUwZ-Vcg5VgqQEiYlRWCbHg0ypY459uA&e=

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

--
Richard

ODL L2 project