Simpler Navigation for Servers and Operating Systems - Please Update Your Bookmarks
Completed: a much simpler Servers and Operating Systems section of the Community. We combined many of the older boards, so you won't have to click through so many levels to get at the information you need. Check the consolidated boards here as many sub-forums are now single boards.
If you have bookmarked forums or discussion boards in Servers and Operating Systems, we suggest you check and update them as needed.
BladeSystem Server Blades
cancel
Showing results for 
Search instead for 
Did you mean: 

554FLB errors post SPP 2014.09.0 firmware update question

chuckk281
Trusted Contributor

554FLB errors post SPP 2014.09.0 firmware update question

A customer question from Hoa:

 

************

 

I have a customer with 460 Gen8 configured w/ 554FLB and 554M using both Redhat &  Oracle UKE in VC FF 10/24 in two enclosures MES BfS configuration.    Post VC 4.31 and SPP 2014.09.0 +MSB  update, they have 17 servers shown up with below errors upon Linux BfS, before firmware update they had two systems behaving this way.      Customer strongly disagree with VC support team’s analysis (uploaded VC support-dump) that this is a Linux issue, not 554FLB OR VC issues.        Has anyone come across this issue before?    

 

webpic346.png

 

**************

 

Reply from Kevin:

 

**********

 

Hello Hoa,

 

It appears that you have the linux bond set for “mode 4” aka 802.3ad which is not supported. Please provide evidence of the bonded script in /etc/sysconfig/network-scripts.

 

************

 

From Hoa:

 

***********

 

Hi Kevin,

 

So what mode Linux bonding VC supports?   That we can recommend.

 

**********

 

From Kevin:

 

************

 

Hello Hoa,

 

Mode 1

 

active_slave

 

        Specifies the new active slave for modes that support it

        (active-backup, balance-alb and balance-tlb).  Possible values

        are the name of any currently enslaved interface, or an empty

        string.  If a name is given, the slave and its link must be up in order

        to be selected as the new active slave.  If an empty string is

        specified, the current active slave is cleared, and a new active

        slave is selected automatically.

 

        Note that this is only available through the sysfs interface. No module

        parameter by this name exists.

 

        The normal value of this option is the name of the currently

        active slave, or the empty string if there is no active slave or

        the current mode does not use an active slave.

 

 

Mode 5

 

balance-tlb or 5

 

               Adaptive transmit load balancing: channel bonding that

               does not require any special switch support.

 

               In tlb_dynamic_lb=1 mode; the outgoing traffic is

               distributed according to the current load (computed

               relative to the speed) on each slave.

 

               In tlb_dynamic_lb=0 mode; the load balancing based on

               current load is disabled and the load is distributed

               only using the hash distribution.

 

               Incoming traffic is received by the current slave.

               If the receiving slave fails, another slave takes over

               the MAC address of the failed receiving slave.

 

               Prerequisite:

 

               Ethtool support in the base drivers for retrieving the

               speed of each slave.

 

 

Mode 6

balance-alb or 6

 

               Adaptive load balancing: includes balance-tlb plus

               receive load balancing (rlb) for IPV4 traffic, and

               does not require any special switch support.  The

               receive load balancing is achieved by ARP negotiation.

               The bonding driver intercepts the ARP Replies sent by

               the local system on their way out and overwrites the

               source hardware address with the unique hardware

               address of one of the slaves in the bond such that

               different peers use different hardware addresses for

               the server.

 

               Receive traffic from connections created by the server

               is also balanced.  When the local system sends an ARP

               Request the bonding driver copies and saves the peer's

               IP information from the ARP packet.  When the ARP

               Reply arrives from the peer, its hardware address is

               retrieved and the bonding driver initiates an ARP

               reply to this peer assigning it to one of the slaves

               in the bond.  A problematic outcome of using ARP

               negotiation for balancing is that each time that an

               ARP request is broadcast it uses the hardware address

               of the bond.  Hence, peers learn the hardware address

               of the bond and the balancing of receive traffic

               collapses to the current slave.  This is handled by

               sending updates (ARP Replies) to all the peers with

               their individually assigned hardware address such that

               the traffic is redistributed.  Receive traffic is also

               redistributed when a new slave is added to the bond

               and when an inactive slave is re-activated.  The

               receive load is distributed sequentially (round robin)

               among the group of highest speed slaves in the bond.

 

               When a link is reconnected or a new slave joins the

               bond the receive traffic is redistributed among all

               active slaves in the bond by initiating ARP Replies

               with the selected MAC address to each of the

               clients. The updelay parameter (detailed below) must

               be set to a value equal or greater than the switch's

               forwarding delay so that the ARP Replies sent to the

               peers will not be blocked by the switch.

 

               Prerequisites:

 

               1. Ethtool support in the base drivers for retrieving

               the speed of each slave.

 

               2. Base driver support for setting the hardware

               address of a device while it is open.  This is

               required so that there will always be one slave in the

               team using the bond hardware address (the

               curr_active_slave) while having a unique hardware

               address for each slave in the bond.  If the

               curr_active_slave fails its hardware address is

               swapped with the new curr_active_slave that was

 

*********

 

From Hoa:

 

**********

 

Do you think this doc should mention VC can’t support Linux bonding mode 4?

 

http://h20566.www2.hp.com/hpsc/doc/public/display?sp4ts.oid=3552695&docId=emr_na-c02957870&lang=en&cc=us

 

***********

 

And from Kevin:

 

************

 

Yes there is a saw article indicating that mode 4 is not supported.

 

http://h20564.www2.hp.com/hpsc/doc/public/display?sp4ts.oid=4144084&docId=emr_na-c03313825-1&docLocale=en_US

 

************

 

Comments?