cancel
Showing results forΒ 
Search instead forΒ 
Did you mean:Β 

setting llt (low prio net) in a bond configured in a bridge

javierv
Level 2

Hi,

I want to set a low prio network over a bond which is part of a bridge. I have a cluster with two nodes.

This is the network configuration (it is the same for both nodes):

node 2:

[root@node2 ]# cat /proc/version
Linux version 2.6.32-504.el6.x86_64 (mockbuild@x86-023.build.eng.bos.redhat.com) (gcc version 4.4.7 20120313 (Red Hat 4.4.7-11) (GCC) ) #1 SMP Tue Sep 16 01:56:35 EDT 2014

 

[root@node2 ]# ifconfig | head -n 24
bond0     Link encap:Ethernet  HWaddr 52:54:00:14:13:21  
          inet6 addr: fe80::5054:ff:fe14:1321/64 Scope:Link
          UP BROADCAST RUNNING MASTER MULTICAST  MTU:1500  Metric:1
          RX packets:761626 errors:0 dropped:0 overruns:0 frame:0
          TX packets:605968 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:1188025449 (1.1 GiB)  TX bytes:582093867 (555.1 MiB)

br0       Link encap:Ethernet  HWaddr 52:54:00:14:13:21  
          inet addr:10.10.11.102  Bcast:10.10.11.255  Mask:255.255.255.0
          inet6 addr: fe80::5054:ff:fe14:1321/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:49678 errors:0 dropped:0 overruns:0 frame:0
          TX packets:50264 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:44061727 (42.0 MiB)  TX bytes:28800387 (27.4 MiB)

eth0      Link encap:Ethernet  HWaddr 52:54:00:14:13:21  
          UP BROADCAST RUNNING PROMISC SLAVE MULTICAST  MTU:1500  Metric:1
          RX packets:761626 errors:0 dropped:0 overruns:0 frame:0
          TX packets:605968 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:1188025449 (1.1 GiB)  TX bytes:582093867 (555.1 MiB)

 

[root@node2 ]# brctl show
bridge name    bridge id        STP enabled    interfaces
br0        8000.525400141321    no        bond0

[root@node2 ]# cat /proc/net/bonding/bond0
Ethernet Channel Bonding Driver: v3.6.0 (September 26, 2009)

Bonding Mode: fault-tolerance (active-backup)
Primary Slave: None
Currently Active Slave: eth0
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 0
Down Delay (ms): 0

Slave Interface: eth0
MII Status: up
Speed: Unknown
Duplex: Unknown
Link Failure Count: 0
Permanent HW addr: 52:54:00:14:13:21
Slave queue ID: 0

 

node 1:

[root@node1]# ifconfig | head -n 24

bond0     Link encap:Ethernet  HWaddr 52:54:00:2E:6D:23 
          inet6 addr: fe80::5054:ff:fe2e:6d23/64 Scope:Link
          UP BROADCAST RUNNING MASTER MULTICAST  MTU:1500  Metric:1
          RX packets:816219 errors:0 dropped:0 overruns:0 frame:0
          TX packets:668207 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:1194971130 (1.1 GiB)  TX bytes:607831273 (579.6 MiB)

br0       Link encap:Ethernet  HWaddr 52:54:00:2E:6D:23  
          inet addr:10.10.11.101  Bcast:10.10.11.255  Mask:255.255.255.0
          inet6 addr: fe80::5054:ff:fe2e:6d23/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:813068 errors:0 dropped:0 overruns:0 frame:0
          TX packets:640374 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:1181039350 (1.0 GiB)  TX bytes:604216197 (576.2 MiB)

eth0      Link encap:Ethernet  HWaddr 52:54:00:2E:6D:23  
          UP BROADCAST RUNNING SLAVE MULTICAST  MTU:1500  Metric:1
          RX packets:816219 errors:0 dropped:0 overruns:0 frame:0
          TX packets:668207 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:1194971130 (1.1 GiB)  TX bytes:607831273 (579.6 MiB)

[root@node1]# brctl show
bridge name    bridge id        STP enabled    interfaces
br0        8000.5254002e6d23    no        bond0

[root@node1 ]# cat /proc/net/bonding/bond0
Ethernet Channel Bonding Driver: v3.6.0 (September 26, 2009)

Bonding Mode: fault-tolerance (active-backup)
Primary Slave: None
Currently Active Slave: eth0
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 0
Down Delay (ms): 0

Slave Interface: eth0
MII Status: up
Speed: Unknown
Duplex: Unknown
Link Failure Count: 0
Permanent HW addr: 52:54:00:2e:6d:23
Slave queue ID: 0

 

the llt configuration files are the following:

[root@node1 ]# cat /etc/llttab
set-node node2
set-cluster 1042
link eth3 eth3-52:54:00:c3:a0:55 - ether - -
link eth2 eth2-52:54:00:35:f6:a5 - ether - -

link-lowpri bond0 bond0 - ether - -

[root@node1 ]# cat /etc/llttab
set-node node1
set-cluster 1042
link eth3 eth3-52:54:00:bc:9b:e5 - ether - -
link eth2 eth2-52:54:00:31:fb:31 - ether - -

link-lowpri bond0 bond0 - ether - -

 

However this seems that is not working. When I check the llt status, the node thinks the interface is down in the other one.

[root@node2 ]# lltstat -nvv | head
LLT node information:
    Node                 State    Link  Status  Address
     0 node1             OPEN    
                                  eth3   UP      52:54:00:BC:9B:E5
                                  eth2   UP      52:54:00:31:FB:31
                                  bond0   DOWN    
   * 1 node2             OPEN    
                                  eth3   UP      52:54:00:C3:A0:55
                                  eth2   UP      52:54:00:35:F6:A5
                                  bond0   UP      52:54:00:14:13:21

[root@node2 ]# lltstat -nvv | head
LLT node information:
    Node                 State    Link  Status  Address
   * 0 node1             OPEN    
                                  eth3   UP      52:54:00:BC:9B:E5
                                  eth2   UP      52:54:00:31:FB:31
                                  bond0   UP      52:54:00:2E:6D:23
     1 node2             OPEN    
                                  eth3   UP      52:54:00:C3:A0:55
                                  eth2   UP      52:54:00:35:F6:A5
                                  bond0   DOWN 

 

Do you know if I have something worng?

Is this a valid configuration?

 

Thanks,

Javier

 

4 REPLIES 4

Gaurav_S
Moderator
Moderator
   VIP    Certified

Hi Javier,

What is the verison of VCS you have on systems ? LLT bond support was introduced from 5.0MP3 onwards

Did you add this dynamically or you have rebooted (or unloaded /reloaded) LLT modules after configuration ?

From the below technote, it appears that you have edited the llttab file correctly however did the LLT modules loaded after config was changed ?

https://www.veritas.com/support/en_US/article.TECH62995

 

G

 

sudhir_h
Level 4
Employee

In addition as mentioned in the TECHNOTE provided by Gaurav, can you confirm if the bonded NIC from all the cluster hosts are connected to the same switch.

Please refer to the Note section in the TECHNOTE 000035027 .

You can configure NIC bonds (aggregated interfaces) as private links under LLT. LLT treats each aggregated interface as a single link. So, you must configure these NICs that form the bond in such a way that the NICs are connected to the same switch or hub.

 

Regards,

Sudhir

javierv
Level 2

Hi,

My VCS version is the following:

[root@node1 litp-admin]# haclus -value EngineVersion
6.1.10.000

 

I did both. Fisrt I used the following commands to set the llt network:

lltconfig -u bond0

lltconfig -t bond0 -d bond0 -l

 

And then I modified the /etc/llttab files and restarted the nodes.

 

After this the interfaces are DOWN in the other nodes.

sudhirh, In this case I only have one ethernet interface attach to the bond, so I think it should not be a problem with the swich.

 

I have found that if I unlink the bond from the bridge:

brctl delif br0 bond0

Then the interfaces are up:

[root@node2 ]# lltstat -nvv|head

LLT node information:
    Node                 State    Link  Status  Address
     0 node1             OPEN
                                  eth3   UP     52:54:00:BC:9B:E5
                                  eth2   UP     52:54:00:31:FB:31
                                  bond0   UP   52:54:00:2E:6D:23
   * 1 node2             OPEN
                                  eth3   UP     52:54:00:C3:A0:55
                                  eth2   UP     52:54:00:35:F6:A5
                                  bond0   UP      52:54:00:14:13:21

[root@node1 ]# lltstat -nvv|head

LLT node information:
    Node                 State    Link  Status  Address
   * 0 node1             OPEN
                                  eth3   UP     52:54:00:BC:9B:E5
                                  eth2   UP     52:54:00:31:FB:31
                                  bond0   UP   52:54:00:2E:6D:23
     1 node2             OPEN
                                  eth3   UP     52:54:00:C3:A0:55
                                  eth2   UP     52:54:00:35:F6:A5
                                  bond0   UP   52:54:00:14:13:21

 

starflyfly
Level 6
Employee Accredited Certified

Hi,

According technote :https://www.veritas.com/support/en_US/article.000035027

and  https://www.veritas.com/support/en_US/article.000015218

bond work as llt is supported.

but :

...

Note: If the NICs are connected to different switches or hubs, you must establish connection between the switches or hubs.
See the Veritas Cluster Server Installation Guide for instructions to configure private heartbeats that use aggregated interfaces.
However, when defining the bond interface in the /etc/llttab file using the mac-address  llt will not work properly. 
Example:
# more /etc/llttab
set-node 10.100.33.135
set-cluster 31337
set-verbose 1
link eth2 eth-00:04:23:c5:d4:78 - ether - -
link eth3 eth-00:04:23:C5:D4:79 - ether - -
link-lowpri bond0 eth-00:14:22:23:27:12 - ether - -
To be more specific when a Linux NIC aggregate interface is configured in the /etc/llttab file as described in Figure 1 'lltstat -nvv | head' will report the link DOWN for all peer nodes. 

..

 

 

if you have modify llttab, bond0 still in down state, you can further check system log.