Emergency UNIX and Linux Support

Hello HP-UX experts,

Server = rx8640 Node partition
OS = HP-UX 11.23
arch = IA64
Network switch = Foundry 16 port switch (1Gbps)

Existing configuration: Tough to explain as it is very messy (see below for the link of zip of network related fles). 2 x 2Gbps aggregates configured some time ago. But suddenly faced some packet loss issue and it was solved by disabling one of the ports (of aggregate) on one of the switch. There were two interfaces which formed lan900 however one cable was going to one switch and another to another switch. Networks team confirmed that both switches were seeing same MAC address and both were receiving packets at the same time hence disabling on of the network port on one of the switch resolved the packet loss issue. however now the server doesn't have any resilience and its running only on one network interface with 1gbps bandwidth. server also has a service guard package name dbXY0.

As it is noticed that 1Gbps throughput would be enough for the server, its decided NOT to use any aggregate instead a simple failover (2 NICs - 1Gbps each active/passive LAN_MONITOR mode). Could anyone please suggest me how to remove existing APA config and configure simple fail over (1Gbps throughput) using LAN_Monitor mode? I'm not entirely sure which Interfaces I need to deconfigure and which interfaces to configure for new setup as service guard config should not get affected.

Pl let me know if any more info needed from my end. Appreciate any professional help!

Thanks a lot.

Note: I'm unable to attach zip of various network commands' output - getting "invalid thread contact admin" error. So uploaded to dropbox, here is the link

Firstly, I'm not a HP-UX expert but having read your post I feel compelled to comment.

In aggregate mode the box should be sending packets on both interfaces (to use the bandwidth) which is the idea of aggregate in the first place.

Any aggregate interfaces on any operating system worth its salt will effectively failover to the surviving interface if one of them fails. So no need to specifically configure 'failover' instead of 'aggregate'.

If the box was sending on both interfaces and experiencing packet loss on one of them then that is a network problem not a Unix problem. Your network boys have changed something. It's quite okay for the network switches to see the same MAC address provided they are configured to expect that.

So, talking purely generically (and not HP-UX specific) I think you should get your network team to fix this issue and not be tempted to mess with your Unix box configuration unnecessarily.

(With your first post going unanswered for 5 days perhaps ask moderators to move it to the HP-UX forum proper where the right experts will see it.)