On 3月30日, 下午4时58分, "AndreasSchulze" <b79...@gmx.de> wrote:
> On 29 Mrz., 18:20, "Hardy" <wyh...@gmail.com> wrote:
>
>
>
> > Hi,
>
> > DB2 HA on two AIX server A and B, HACMP take over test is OK.
> > but when we issue"halt -q" on on server A, B can take over all the
> > resource of A but it is very slow when it come to "db2start", 16nodes
> > start cost 30 minutes। but when started, it works well.
>
> > some problems about networking is reported:
> > 1. the service ip of B is moved to another interface. also the service
> > ip of A is moved to the same interface.
> > 2। "Interface 192.168.7.3 has failed on node PDBA","Interface
> > 192.168.7.3 is now available on node PDBA". 7.3 is bootip of B.
>
> > for 1, why service ip for B is moved? is it normal?
> > for 2, the reason? what it is doing?
>
> > Thank you.
>
> Hallo Hardy
>
> Too little information for indepth analysis. Some thoughts:
> Depending on HACMP release service IP might move if the cluster
> detects network down on the originating adapter (resulting from
> halting node A? How many interfaces are reachable on this particular
> network?). This would explain that the service IP from A is placed on
> the other adapter also.
> Depending on vg type starting the DB might take considerably longer
> after a halt -q because of fsck and/or replaying logs (both
> filesystems and DB logs)?
>
> As to your 2nd problem: that looks like some sort of misconfiguration
> (works as designed). Check AIX and HACMP configuration on both nodes
> for information about 192.168.7.3.
>
> hth
> andreas

Thank you, Andreas. It's DB2 cluster, all cascating vgs, NFS used for
db2 instance home.
the time for disk checking, vg takeover, filesystem log replaying is
short and successful.
vg,fs,db2 cfg, nfs fs all are OK. only db2start ver slow.

now I'm considering the IY92334 and some tests are planned to do.

Thank you.