flaky virtual IP for MC/Service guard

[Frank_9]

We run a cluster using MC/Service Guard. The cluster is composed of two nodes .2 and .3 with virtual IP .1
On a local lan all interfaces respond to PING, from two routers away only .2 and .3 respond, .1 does not answer. We are using Cisco routers and when we clear the ARP cache on the router local to the cluster all works for a few minutes (from seconds to 25 minutes) then the virtual IP refused to respond. Other machines local to the MC/Service Guard cluster respond to PING from two routers away as well as the physical IPs .2 and .3 The virtual IP .1 is the only problem.

In the configuration of MC/Service Guard, should a virtual MAC address be defined? Has anyone else crossed this problem? This is the only problem on our network and it is only the virtual IP that refuses to respond, though the problem does not occur if the virtual IP is PINGed from the local lan to the clsuer.

THanks,

Frank

[Ovidiu D. Raita]

Frank,

You don't to set any "virtual MAC" address. The virtual IP should pass the routers without any problem. As far as I know there's no special setting needed at the UNIX and/or router level.

Make sure
- you run a HP supported MC Service Guard configuration i.e you have the two systems .1 and .2 on the same subnet and they are not separated by an active networking device(router).
- you don't have another system with the same IP somewhere in the net (probably behind the router). This can be reason why you can access the addres from inside the cell and for a while from outside too, but as soon as the other system starts sending it's MAC the ARP tables get changed. IF you have a chance have a look at the ARP tables from the router and verify if all network devices support virtual IPs.

Let me know if I'm wrong,
Ovidiu

[Dieter Degrendele_1]

Hi,

I had once the same problem. Some subnets could reach the machine, others not. After weeks of investigations, I suggested to shutdown the switch (we were using ALCATLEL and CISCO). Problem was solved immediately. I have the impression that it has to do something with the ARP cache.

Rgds,
DD

[Brian Hackley]

Hi,
Hmmm this sounds familiar...there was a problem reported 2 years ago where the re-arp
broadcast was broken, but that was fixed in an 11.0 ARPA Transport patch. Check to see that you have PHNE_17662 or later. The SR # is 1653282012.
Hope this helps,
-> Brian Hackley

[Gengwd]

Hi all,

I have the same sense of Frank. I'm meeting the same problem and I don't know how to solve it.

Is the way of "shutdown switch" the permanent way for solving this problem?

If no, can installation of HP-UX network patch works? and is there any risk for installing the HP-UX network patch?

My enviroment:
HP-UX 11.0
MC/ServiceGuard 11.09
No MC/SG patches installed

[melvyn burnard]

You MUST have all the latest networking patches installed, and the patch for MC/SG 11.09 which is currently PHSS_22876

[Gengwd]

If I installed MC/SG patch PHSS_22876,is it necessary to install networking patch such as PHNE_22397?

Thanks for melvyn's great help.

[melvyn burnard]

Yes the network patch should be installed.
Having said that , I guess it is time I posted a list of recommended patches for SG nodes on HP-UX 11.00 :-}
I will do this as a separate new message

[melvyn burnard]

Ok, just for the record, here is te RECOMMENDED patch list for SG on 11.00

Please note that any or all of these patches could be superseded, and could have dependencies that you will have to ensure are included.
Also, various patches will/will not apply depending on the software used on your system.



PATCH Number Title

PHCO_21901 s700_800 11.00 fuser(1M) cumulative patch
PHCO_22733 s700_800 11.00 sar(1M) cumulative patch
PHCO_22839 s700_800 11.00 LVM commands cumulative patch
PHKL_13431 s700_800 11.00 bad block msg,panic; data possibly corrupted
PHKL_18543 s700_800 11.00 PM/VM/UFS/async/scsi/io/DMAPI/JFS/perf patch
PHKL_18800 s700_800 11.00 Cumulative JFS patch - panic:dirty inval
PHKL_21834 s700_800 11.00 Fibre Channel Mass Storage Driver
PHKL_22494 s700_800 11.00 PCI ExpROM,bridge,BA hints,Lowfat,PIOP,EPIC
PHKL_22759 s700_800 11.00 SCSI IO Subsystem Cumulative Patch
PHKL_22986 s700_800 11.00 cumulative pstat patch
PHKL_23127 s700_800 11.00 LVM Cumulative patch
PHNE_15537 s700_800 11.00 Fibre Channel cumulative patch
PHNE_16172 s700_800 11.00 gated (1M) patch
PHNE_18635 s700_800 11.00 HP Auto-Port Aggregation patch B.11.00.02
PHNE_19200 s700_800 11.00 EISA 100BT patch
PHNE_19633 s700_800 11.00 HP-PB FDDI (J2157B) product cumulative patch
PHNE_20874 s700_800 11.00 PCI FDDI B.11.00.[01-03,05,08,11-13] patch
PHNE_20949 s700_800 11.00 HyperFabric B.11.00.0[0-8] cumulative patch
PHNE_21389 s700_800 11.00 EISA FDDI B.11.00.0[0-7] cumulative patch
PHNE_21390 s700_800 11.00 HSC FDDI B.11.00.0[0-7] cumulative patch
PHNE_21731 s700_800 11.00 r-commands cumulative mega-patch
PHNE_22244 s700_800 11.00 Built-in PCI 100BASE-T patch
PHNE_22245 s700_800 11.00 PCI 100B(Tx/Fx) cumulative patch
PHNE_22353 s700_800 11.00 LAN product cumulative patch
PHNE_22397 s700_800 11.00 cumulative ARPA Transport patch
PHNE_22566 s700_800 11.00 Cumulative STREAMS Patch
PHNE_22728 s700_800 11.00 4 port PCI 100BASE-T cumulative patch
PHNE_22930 s700_800 11.00 1000Base-SX/T B.11.00.[02-11] cumul. patch
PHSS_17496 s700_800 11.00 Predictive C.11.0[0,a-m] cumulative patch
PHSS_22946 s700_800 11.00 X/Motif 32bit Runtime JAN2001 Periodic Patch