Trying to create an external bridge

[Xavier_P_]

Hi guys,

I'm trying to create an external bridge following the users guide from Intel.

First of all I execute the following command:

micctrl --addbridge=br0 --type=external --ip=<ip_HOST>

Then, I can see that a file ifcfg-br0 has created and I can see the bridge created with brctl show. Also I attached the eth1 interface to the bridge br0

Then, when I try execute:

micctrl --network=static --bridge=br0 --ip=<ip_MIC0>

I always obtain this error:

[Error] Bridge 'br0' not defined

Can you help me?

Thanks in advance,

Xavi.

[Sunny_G_Intel]

Hi Xavi,

Can you please refer to the section 3.1.3 External Bridge Topology in the latest Intel® Xeon Phi™ System Administration guide. In the list of steps above, I see that you have not updated /etc/sysconfig/network-scripts/ifcfg-eth0 file as required. Can you please verify if you have already done that.

Thanks

[Xavier_P_]

Hi Sunny,

thanks for you quick reply. In my case I've attached the eth1 interface to the br0 bridge and I have this inside the /etc/sysconfig/network-scripts/ifcfg-eth1 file:

DEVICE=eth1
ONBOOT=yes
TYPE="ethernet"
BRIDGE=br0
NM_CONTROLLED="no"

I forgot to include this in the steps.

Thanks,

Xavi.

[Sunny_G_Intel]

Hi Xavi, 

Can you please verify that the network settings were updated. Check if eth1 is already tied to br0 (verify this using ifconfig eth1 and ifconfig br0). 

service network restart  //This should  bring up br0 without errors

Also can you please share output of 

 micctrl --config mic0

Thanks

[Xavier_P_]

Hi Sunny,

this is the output of ifconfig:

ifconfig eth1
eth1      Link encap:Ethernet  HWaddr 00:25:90:E2:30:25  
          inet addr:10.2.1.1  Bcast:10.2.255.255  Mask:255.255.0.0
          inet6 addr: fe80::225:90ff:fee2:3025/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:137626 errors:0 dropped:0 overruns:0 frame:0
          TX packets:64173 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:11163672 (10.6 MiB)  TX bytes:4211219 (4.0 MiB)
          Memory:fbe00000-fbe20000

ifconfig br0
br0       Link encap:Ethernet  HWaddr 00:25:90:E2:30:25  
          inet addr:192.168.16.200  Bcast:192.168.16.255  Mask:255.255.255.0
          inet6 addr: fe80::225:90ff:fee2:3025/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:322 errors:0 dropped:0 overruns:0 frame:0
          TX packets:146 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:22211 (21.6 KiB)  TX bytes:10909 (10.6 KiB)

I don't know why eth1 have this IP address. I can see the eth1 tied to br0 through brctl

brctl show
bridge name    bridge id        STP enabled    interfaces
br0        8000.002590e23025    no        eth1

The output of micctrl shows that mic0 it isn't configured

micctrl --config mic0

mic0:
=============================================================
  [Error] Not configured

I don't understand...

Thanks

[Xavier_P_]

I think there are more things wrong.

I've installed mpss 3.4.3 and before that, I've uninstalled old versions, but now I can't find the file /lib/firmware/mic/uos.img.

And many commands like micctrl --boot fails with the output [Warning] mic0: Parse configuration failed - skipping

I don't what to do.

Xavi.

[Sunny_G_Intel]

Hello Xavi,

Before setting up an external bridge, I would recommend verifying the basic Intel® MPSS installation. Can you please execute the following commands.

//Stop the MPSS daemon
service mpss stop

//Remove any external bridge you created for MIC
micctrl --network=default
micctrl --delbridge=br0

//Delete ifcfg-br0 file if it was created by micctrl i.e. used for nothing else besides external bridge

//Reset to initial settings
micctrl --cleanconfig
micctrl --initdefaults

//Verify some of the mpss config files 
/etc/mpss/*conf 
/var/mpss/mic0/etc/network/interfaces

//If the files look OK then start the MPSS daemon
service mpss start

//Run miccheck
miccheck

//See if you can ping the coprocessor and Enable passwordless ssh to login to the coprocessor

Once you have the MPSS configured correctly with default settings you could try setting up external bridge as follows

//Modify the /etc/sysconfig/network-scripts/ifcfg-eth1 file as follows
DEVICE=eth1
NM_CONTROLLED=no
TYPE=Ethernet
ONBOOT=yes
BRIDGE=br0

//Add the bridge -- assuming sample ip address for host 
micctrl --addbridge=br0 --type=external --ip=10.10.10.2

//Check if ifcfg-br0 was created in /etc/sysconfig/network-scripts directory

//Restart the network 
service network restart

//Attach the coprocessors to the bridge you just created 
micctrl --network=static --bridge=br0 --ip=10.10.10.3

//Now log in to the coprocessor and check if you can ping outside your host on the same network. For eg. ping different node of the same cluster

Thanks

[Frances_R_Intel]

As Sunny points out, you probably want to go through the whole initial setup. You don't say what release you were coming from but things have moved around between 2.1 and 3.0 and between 3.2 and 3.3. If you want to incorporate some of your older configuration information or additional files you are including in the root file system on the coprocessor, the new location of the micX.conf files is /etc/mpss, the new location for the files for the coprocessor root file system (except the base image) are in /var/mpss.

[Xavier_P_]

Hi, thanks for your replies.

Finally I can boot the mic and connect without configuring the bridge, it was a problem of the configuration file. Now I'm trying to create an external bridge again.

If I follow the instructions and execute

micctrl --network=static --bridge=br0 --ip=10.10.10.3

I obtain the same error as before: [Error] Bridge 'br0' not defined

The output of micctrl --config is:

micctrl --config

mic0:
=============================================================
    Config Version: 1.1

    Linux Kernel:   /usr/share/mpss/boot/bzImage-knightscorner
    Map File:       /usr/share/mpss/boot/System.map-knightscorner
    BootOnStart:    Enabled
    Shutdowntimeout: 300 seconds

    ExtraCommandLine: 'highres=off'
    PowerManagment: cpufreq_on;corec6_on;pc3_on;pc6_on

    Root Device:   Dynamic Ram Filesystem /var/mpss/mic0.image.gz from:
    Base:      CPIO /usr/share/mpss/boot/initramfs-knightscorner.cpio.gz
    CommonDir: Directory /var/mpss/common
    Micdir:    Directory /var/mpss/mic0
mic0: Unknown network configuration type
        MIC MAC:   Not Configured
        Host MAC:  Not Configured

    LDAP:          Disabled
     NIS:          Disabled

    Cgroup:
        Memory:    Disabled

    Console:        hvc0
    VerboseLogging: Disabled
    CrashDump:      /var/crash/mic 16GB

It seams that there's something wrong with the network configuraton

On the other hand I've created a br0 with brctl and attached to eth1 and mic0 and now I can ping another node different form the host. I think it's done but I would like to know what causes the error "unknown network configuration type".

Thanks a lot for yout time.

Xavi.

[Xavier_P_]

Hi again, all of this is like a maze...

Now I'm trying to configure LDAP inside mic but it happens a very weird thing.

I have a backup of the mic0.conf.

When I run micctrl --rpmdir='/root/mpss-3.4.3/ldap_mic/' -vv the output is:

 [Info] mic0: [Parse] /etc/mpss/mic0.conf
   [Info] mic0: [Parse] Configuration version 1.1
   [Info] mic0: [Parse] /etc/mpss/default.conf
[Filesys] mic0: Update K1omRpms in /etc/mpss/mic0.conf

After that mic0.conf file disappears and in the next I don't know what to do again...

this is the output:

micctrl --ldap=192.168.17.3 --base=ou=cap,ou=cesca,o=root
[Warning] mic0: Not configured - skipping

I think the micctrl is deleting mic0.conf every time...

Xavi.

[Sunny_G_Intel]

Hi Xavier,

Can you please share coprocessor configuration files from /etc/mpss/ directory. 

Thanks

[Xavier_P_]

Hi Sunny,

here it is the mic0.conf

Version 1 1
Include default.conf
Include conf.d/*.conf
OSimage /usr/share/mpss/boot/bzImage-knightscorner /usr/share/mpss/boot/System.map-knightscorner
BootOnStart enabled
PowerManagement cpufreq_on;corec6_on;pc3_on;pc6_on
ExtraCommandLine 'highres=off'
Cgroup memory=disabled
ShutdownTimeout 300
Base CPIO /usr/share/mpss/boot/initramfs-knightscorner.cpio.gz
RootDevice RAMFS /var/mpss/mic0.image.gz
CommonDir /var/mpss/common
MicDir /var/mpss/mic0
Hostname collserola1-mic0
MacAddrs Serial

ExtraCommandLine "highres=off"

# MIC Console
Console "hvc0"

Thanks for your time.

Xavi.