add support for the IVS virtual switch #159

rlane · 2013-06-27T21:54:55Z

IVS is an open source virtual switch available for download at
https://github.com/floodlight/ivs. It uses the openvswitch
kernel module.

I also modified test_hifi slightly to let me run those tests against IVS.

IVS is an open source virtual switch available for download at https://github.com/floodlight/ivs. It uses the openvswitch kernel module.

lantz · 2013-06-27T23:07:33Z

This is interesting - IVS is under my radar, although perhaps it was part of the Switch Light announcement?

Could you explain what the advantage of IVS is over OVS?

One thing I like about OVS is that there are Ubuntu packages for it - are there ubuntu packages for IVS?

All of the other switch classes in Mininet proper also include support to install the switch in install.sh.

One other thing which isn't obvious to me: should Mininet include support for all extant OF software switches (including OVS, Stanford 1.0 switch, CPqD switch, LINC, pyswitch, etc.) or should they be separate features which people can download and install?

One of the GSoC projects involves creating a repository for user-contributed enhancements, topologies, switch classes, apps, etc. for Mininet, and perhaps full system examples as well.

rlane · 2013-06-28T00:26:10Z

One nice feature of IVS is the ability to spread processing of kernel flowtable misses across multiple threads.

The IVS repository can build Debian packages but they are not yet included in any distributions. I'll add support to install.sh to clone the IVS repo and install it. I'll let you know when I've pushed this and the dpctl support to this pull request.

I'm not sure I see a benefit in moving switch classes out of the mininet repository. You'd need to carefully manage the Switch interface (and any core mininet APIs switch classes use) to ensure that it stays compatible with switch classes in other repositories. Given how little code a switch class is it seems better to keep them with the rest of mininet. I wouldn't mind if each switch class were moved to its own file, though.

dpctl is not included with IVS. The user will need to obtain it from the OpenFlow reference repository.

rlane · 2013-06-29T15:54:18Z

I've pushed a couple of commits adding support for dpctl and installing IVS from install.sh. I tested the install script on the mininet VM.

One note on dpctl - we don't have our own version, so users will need to take it from the OpenFlow reference repository.

lantz · 2013-06-29T22:20:32Z

I recommend using ovs-ofctl, which should work hopefully!

rlane · 2013-06-30T01:13:56Z

Ok, I changed the dpctl call to use ovs-ofctl.

IVS works fine using --innamespace.

rlane · 2013-07-08T21:17:01Z

Good to merge?

lantz · 2013-07-09T02:07:52Z

What is the easiest way to verify that it passes the tests (or at least test_hifi)?
Can we do this in an automated fashion?
I want to make sure that we can test all of the switches and make sure they all pass all of the tests.

lantz · 2013-07-09T02:42:41Z

My tests seem to fail when OVS is running - what can we do to allow IVS and OVS to peacefully coexist?

lantz · 2013-07-09T02:44:06Z

I also notice that mn -c doesn't seem to work with IVS.

lantz · 2013-07-09T02:46:05Z

The installation of IVS doesn't seem to install any man pages, notably a man page for for ivs-ctl seems to be missing.

This could happen if OVS was installed but not running.

rlane · 2013-07-09T16:13:14Z

I pushed a couple of commits fixing mn -c. The main problem I saw was ovs-vsctl waiting forever to connect to ovsdb.

For test_hifi I set SWITCH to IVSSwitch to run the test. Maybe SWITCH could be set based on an environment variable. For test_nets I just added IVSSwitch to the SWITCHES dictionary. I'd also suggest removing that dictionary and replacing it with an environment variable choosing a single switch.

I think the only thing we can do to make IVS and OVS coexist is to run them in separate namespaces. Otherwise OVS will delete IVS's kernel datapaths as soon as they're created.

After looking at it more I don't think --innamespace actually works. Shouldn't it be running the switches under mnexec? Since I added the listen connection support for dpctl, now --innamespace is broken with IVS because all the switches are trying to listen on port 6634.

The man pages are currently part of the debian packaging, so won't be included in a manual install. We're going to fix that in the future as we package for more distros.

lantz · 2013-07-09T20:58:36Z

It's important to me that I can issue a single command to confirm that IVS works without having to edit a source file.
It should not be a problem that all of the switches are listening on port 6634 if they're in separate namespaces.
Maybe I should look at this as well but I think it's important that we can come up with an example that uses both IVS and OVS in the same network. Not sure if this is a blocking issue on the merge though - for now we can probably just say use one or the other but not both simultaneously.
That being said, it's important for users to easily be able to figure out how to start up an IVS-based Mininet even when OVS is running. The reason is that OVS is currently the default switch and is automatically installed when you install Mininet. Maybe it's OK for IVS to complain and say that you need to stop OVS?

Each switch gets its own class so that the test results are listed separately.

rlane · 2013-07-10T19:24:03Z

I pushed a commit that adds IVS to the test suite. It also breaks the tests up so each switch gets its own set of tests, so it's easier to tell if the test failed for a particular switch. The userspace tests seem to have a race condition between starting the switch and running the test, so they fail for me unless I insert some time.sleep(1) calls (not included in the patch).
Agreed, but I don't think the switches are actually being placed in separate namespaces. How is this supposed to currently work for the userspace switch?
Yeah, until we can get switch namespaces to work we'd have to say not to use both simultaneously.
From what I can tell it's actually OK to have the ovs-vswitchd daemon running while using IVS for mininet. The problems occur when OVS is restarted or (maybe) reconfigured, at which point it deletes any datapaths it doesn't own. Does this match what you saw when you tried to run mininet with IVS?

lantz · 2013-07-12T22:59:00Z

I think we should be able to get switch namespaces to work. If IVS works in a namespace, then OVS can work in the root namespace. It may also be possible to get OVS to work in a namespace.

It looks like things are somewhat better now; I was able to run

for s in user ovsk ivs; do sudo mn --switch $s --test pingall; done

successfully. (Note that there is currently a race condition since the pingall test doesn't wait for all switches to connect.)

lantz · 2013-07-12T22:59:33Z

For --switch user --innamespace, we create a separate routed control network and place all switches in their own namespaces.

# mn --switch user --innamespace --topo tree,2,2 -v output
s1 -> c0 
c0 -> s1 
*** Results: 0% dropped (0/2 lost)
s2 -> c0 
c0 -> s2 
*** Results: 0% dropped (0/2 lost)
s3 -> c0 
c0 -> s3 
*** Results: 0% dropped (0/2 lost)
mininet> pingall
*** Ping: testing ping reachability
h1 -> h2 h3 h4 
h2 -> h1 h3 h4 
h3 -> h1 h2 h4 
h4 -> h1 h2 h3 
*** Results: 0% dropped (0/12 lost)
mininet> s1 ifconfig
lo        Link encap:Local Loopback  
          inet addr:127.0.0.1  Mask:255.0.0.0
          UP LOOPBACK RUNNING  MTU:16436  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:0 (0.0 B)  TX bytes:0 (0.0 B)

s1-eth0   Link encap:Ethernet  HWaddr 66:3e:38:47:6a:68  
          inet addr:192.168.123.2  Bcast:192.168.255.255  Mask:255.255.0.0
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:36 errors:0 dropped:0 overruns:0 frame:0
          TX packets:37 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:4556 (4.5 KB)  TX bytes:5310 (5.3 KB)

s1-eth1   Link encap:Ethernet  HWaddr 46:4e:ed:56:a2:64  
          UP BROADCAST RUNNING PROMISC MULTICAST  MTU:1500  Metric:1
          RX packets:13 errors:0 dropped:0 overruns:0 frame:0
          TX packets:13 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:1084 (1.0 KB)  TX bytes:1084 (1.0 KB)

s1-eth2   Link encap:Ethernet  HWaddr 62:a8:7c:e4:66:88  
          UP BROADCAST RUNNING PROMISC MULTICAST  MTU:1500  Metric:1
          RX packets:13 errors:0 dropped:0 overruns:0 frame:0
          TX packets:13 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:1084 (1.0 KB)  TX bytes:1084 (1.0 KB)

tap0      Link encap:Ethernet  HWaddr 00:00:00:00:00:01  
          UP BROADCAST RUNNING PROMISC MULTICAST  MTU:1500  Metric:1
          RX packets:6 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:500 
          RX bytes:360 (360.0 B)  TX bytes:0 (0.0 B)

mininet>

IVS needs to be sent SIGTERM so it has a chance to clean up the kernel datapath.

This change uses the `Node.cmd` method instead of `Popen`. The `cmd` method sends the input to a shell which may be in another namespace (if --innamespace is in use), while `Popen` would always run in the root namespace.

rlane · 2013-07-19T01:16:58Z

I added a commit to support running IVS with --innamespace. Here's the output from ivs-ctl showing that each switch node only sees one datapath:

mininet> s1 ivs-ctl show
s1:
  kernel lookups: hit=0 missed=58 lost=0
  kernel flows=0
  ports:
    0 s1 (internal)
      rx: packets=0 bytes=0 errors=0 dropped=0
      tx: packets=0 bytes=0 errors=0 dropped=0
    1 s1-eth1 
      rx: packets=29 bytes=1858 errors=0 dropped=0
      tx: packets=29 bytes=1858 errors=0 dropped=0
    2 s1-eth2 
      rx: packets=29 bytes=1858 errors=0 dropped=0
      tx: packets=29 bytes=1858 errors=0 dropped=0
mininet> s2 ivs-ctl show
s2:
  kernel lookups: hit=0 missed=53 lost=0
  kernel flows=0
  ports:
    0 s2 (internal)
      rx: packets=0 bytes=0 errors=0 dropped=0
      tx: packets=0 bytes=0 errors=0 dropped=0
    1 s2-eth1 
      rx: packets=12 bytes=840 errors=0 dropped=0
      tx: packets=27 bytes=1662 errors=0 dropped=0
    2 s2-eth2 
      rx: packets=12 bytes=840 errors=0 dropped=0
      tx: packets=27 bytes=1662 errors=0 dropped=0
    3 s2-eth3 
      rx: packets=29 bytes=1858 errors=0 dropped=0
      tx: packets=29 bytes=1858 errors=0 dropped=0
mininet> s3 ivs-ctl show
s3:
  kernel lookups: hit=0 missed=53 lost=0
  kernel flows=0
  ports:
    0 s3 (internal)
      rx: packets=0 bytes=0 errors=0 dropped=0
      tx: packets=0 bytes=0 errors=0 dropped=0
    1 s3-eth1 
      rx: packets=12 bytes=840 errors=0 dropped=0
      tx: packets=27 bytes=1662 errors=0 dropped=0
    2 s3-eth2 
      rx: packets=12 bytes=840 errors=0 dropped=0
      tx: packets=27 bytes=1662 errors=0 dropped=0
    3 s3-eth3 
      rx: packets=29 bytes=1858 errors=0 dropped=0
      tx: packets=29 bytes=1858 errors=0 dropped=0

The root namespace sees no datapaths.

lantz · 2013-07-27T01:44:42Z

I'm out of town at the moment and also rather sick unfortunately, but that sounds good - I will take a look at it when I get back next week.

lantz · 2013-08-01T00:50:56Z

Reopening to fix commit history. ;-p

rlane added 2 commits June 27, 2013 14:50

add support for the IVS virtual switch

27da832

IVS is an open source virtual switch available for download at https://github.com/floodlight/ivs. It uses the openvswitch kernel module.

test_hifi: use SWITCH to pick the switch class

812c91c

rlane added 2 commits June 27, 2013 17:56

IVSSwitch: add support for dpctl

91261b2

dpctl is not included with IVS. The user will need to obtain it from the OpenFlow reference repository.

install.sh: add support for IVS

8ee4aa6

IVSSwitch: use ovs-ofctl for dpctl functionality

803c0a6

IVSSwitch: remove namespace warnings

71ffb00

IVS works fine using --innamespace.

rlane added 2 commits July 9, 2013 08:34

clean: don't wait forever for ovsdb

94ff77f

This could happen if OVS was installed but not running.

clean: kill ivs processes

60abb34

tests: run all tests with OVS, IVS, and userspace switches

0e2cc60

Each switch gets its own class so that the test results are listed separately.

ghost assigned lantz Jul 12, 2013

rlane added 2 commits July 18, 2013 13:50

clean: send SIGTERM before SIGKILL

5c24263

IVS needs to be sent SIGTERM so it has a chance to clean up the kernel datapath.

IVSSwitch: support running IVS in a namespace

0a54360

This change uses the `Node.cmd` method instead of `Popen`. The `cmd` method sends the input to a shell which may be in another namespace (if --innamespace is in use), while `Popen` would always run in the root namespace.

lantz closed this in e8d6ebe Aug 1, 2013

lantz reopened this Aug 1, 2013

lantz closed this in 786117d Aug 1, 2013

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

add support for the IVS virtual switch #159

add support for the IVS virtual switch #159

rlane commented Jun 27, 2013

lantz commented Jun 27, 2013

rlane commented Jun 28, 2013

rlane commented Jun 29, 2013

lantz commented Jun 29, 2013

rlane commented Jun 30, 2013

rlane commented Jul 8, 2013

lantz commented Jul 9, 2013

lantz commented Jul 9, 2013

lantz commented Jul 9, 2013

lantz commented Jul 9, 2013

rlane commented Jul 9, 2013

lantz commented Jul 9, 2013

rlane commented Jul 10, 2013

lantz commented Jul 12, 2013

lantz commented Jul 12, 2013

rlane commented Jul 19, 2013

lantz commented Jul 27, 2013

lantz commented Aug 1, 2013

add support for the IVS virtual switch #159

add support for the IVS virtual switch #159

Conversation

rlane commented Jun 27, 2013

lantz commented Jun 27, 2013

rlane commented Jun 28, 2013

rlane commented Jun 29, 2013

lantz commented Jun 29, 2013

rlane commented Jun 30, 2013

rlane commented Jul 8, 2013

lantz commented Jul 9, 2013

lantz commented Jul 9, 2013

lantz commented Jul 9, 2013

lantz commented Jul 9, 2013

rlane commented Jul 9, 2013

lantz commented Jul 9, 2013

rlane commented Jul 10, 2013

lantz commented Jul 12, 2013

lantz commented Jul 12, 2013

rlane commented Jul 19, 2013

lantz commented Jul 27, 2013

lantz commented Aug 1, 2013