hacmp problem

hacmp problem

Post by R » Fri, 01 Feb 2002 05:59:21



After I added a cluster, two nodes, and two adapters, I ran
"Synchronize Cluster Topology" and got the following error message:
Connection to remote host refused.

Both hosts' IP addresses and names were included in /etc/hosts and
/.rhosts files. Anyone has a clue?

RR

 
 
 

hacmp problem

Post by Miljenko Jandr » Fri, 01 Feb 2002 13:25:37



Quote:>After I added a cluster, two nodes, and two adapters, I ran
>"Synchronize Cluster Topology" and got the following error message:
>Connection to remote host refused.

>Both hosts' IP addresses and names were included in /etc/hosts and
>/.rhosts files. Anyone has a clue?

First the simplest one: .rhosts muste have permissions of 600, or it
is ignored.

Next, are you sure systems are using /etc/hosts first? Check
/etc/netsvc.conf. Do the systems allow you to do "rsh" as root between
them?

mj

 
 
 

hacmp problem

Post by Matthew Land » Sat, 02 Feb 2002 03:03:58



> After I added a cluster, two nodes, and two adapters, I ran
> "Synchronize Cluster Topology" and got the following error message:
> Connection to remote host refused.

> Both hosts' IP addresses and names were included in /etc/hosts and
> /.rhosts files. Anyone has a clue?

> RR

1) Check that your /.rhosts is permission 600.

2) If that doensn't help, reboot the servers.  After updating /.rhosts
   for some reason HACMP needs a reboot.  

 - Matt
--
_______________________________________________________________________

  IBM High Speed Interconnect - Fibre Channel I/O Dev/Test/Support
   << Comments, views, and opinions are mine alone, not IBM's. >>

 
 
 

hacmp problem

Post by Simon Marches » Sat, 02 Feb 2002 18:48:14




> > After I added a cluster, two nodes, and two adapters, I ran
> > "Synchronize Cluster Topology" and got the following error message:
> > Connection to remote host refused.

> > Both hosts' IP addresses and names were included in /etc/hosts and
> > /.rhosts files. Anyone has a clue?

> > RR

> 1) Check that your /.rhosts is permission 600.

> 2) If that doensn't help, reboot the servers.  After updating /.rhosts
>    for some reason HACMP needs a reboot.

>  - Matt
> --
> _______________________________________________________________________

>   IBM High Speed Interconnect - Fibre Channel I/O Dev/Test/Support
>    << Comments, views, and opinions are mine alone, not IBM's. >>

The reboot is usually required because HACMP adds the "godm" service (used
to retrieve remote node configs) to /etc/services but does not "refresh -s
inetd". Pity.
 
 
 

hacmp problem

Post by Matthew Land » Sun, 03 Feb 2002 05:50:20



> The reboot is usually required because HACMP adds the "godm" service (used
> to retrieve remote node configs) to /etc/services but does not "refresh -s
> inetd". Pity.

Does this mean you can avoid the reboot by just refressing inetd?  If
so, I am going to log a defect.  :)

  - Matt
--
_______________________________________________________________________

  IBM High Speed Interconnect - Fibre Channel I/O Dev/Test/Support
   << Comments, views, and opinions are mine alone, not IBM's. >>

 
 
 

hacmp problem

Post by Simon Marches » Tue, 05 Feb 2002 02:09:39




> > The reboot is usually required because HACMP adds the "godm" service (used
> > to retrieve remote node configs) to /etc/services but does not "refresh -s
> > inetd". Pity.

> Does this mean you can avoid the reboot by just refressing inetd?  If
> so, I am going to log a defect.  :)

Please do! Ironically, IBM pre-sales is the last place development ever seem to
listen to regarding improving product.
 
 
 

hacmp problem

Post by tushen » Sat, 06 Apr 2002 16:20:16


help! hacmp/es+ 4.3.3 ML08 , rlogin or rcp rsh to the other host no
problem and .rhosts is 600 and have reboot the two box when I sync the
topology ,the message is "connection to remote host refused", I have
found that one of the nodes have problem with the "cluster security"
menu,i cannot enter it !!!Anyone has a clue?

--
ts

Posted via dBforums
http://dbforums.com

 
 
 

hacmp problem

Post by Dan Fost » Sat, 06 Apr 2002 20:05:17




>help! hacmp/es+ 4.3.3 ML08 , rlogin or rcp rsh to the other host no
>problem and .rhosts is 600 and have reboot the two box when I sync the
>topology ,the message is "connection to remote host refused", I have
>found that one of the nodes have problem with the "cluster security"
>menu,i cannot enter it !!!Anyone has a clue?

SMIT menus are stored in ODM. That plus the fact you're getting
'connection to remote host refused' suggests that both nodes aren't in
sync for its cluster settings.

Make sure that rlogind and rshd is enabled on both nodes in inetd.conf,
and do a 'refresh -s inetd' then a cluster sync. (You may want to consider
stopping HACMP on both nodes to do a sync of the topology *then* of the
resources.)

-Dan

 
 
 

hacmp problem

Post by Douglas R. Probs » Sun, 07 Apr 2002 11:13:24


Yes, we have this problem or rsh not working as expected on a P690 in LPAR
mode with hacmpes.
If you put all your interfaces in the .rhost on both servers it should work.
This is not what you would expect to have to do, but it will get you around
the problem.  We are still waiting for IBM to get us a fix for this.
Doug


Quote:> help! hacmp/es+ 4.3.3 ML08 , rlogin or rcp rsh to the other host no
> problem and .rhosts is 600 and have reboot the two box when I sync the
> topology ,the message is "connection to remote host refused", I have
> found that one of the nodes have problem with the "cluster security"
> menu,i cannot enter it !!!Anyone has a clue?

> --
> ts

> Posted via dBforums
> http://dbforums.com

 
 
 

1. HACMP problems

Hi everyone, have a couple of odd problems I was hoping to find some
answers to.  We are using older version of HA 4.3.1 on aix 4.3.3
(patches to 06) on
two H80's (cascading.)  Anyhow we used to have a lot of dms timeouts
but everything has been fine for last 6 months - we had another one on
Monday though.  Bad part is the secondary server all
filesystems/programs came up but the tcpip hostname/ip didn't
work..any hints?  We are planning to upgrade soon is it easy?  Also
has anyone seen this;

*** ADDN wms01 10.1.66.141 (noHb17514266) ***
*** ADDN wms01 10.1.30.171 (noHb17514267) ***
giving up on message (100 1 4 0 3 -25119652) NEW EVENT to wms01 (no
retries - )
May  6 15:16:05 EVENT START: node_down wms01

What does that stuff mean? (from cm.log)  or where can I find more
info about where dms error came from?

Detail Data
DUMP STATUS
LED:700
csa:00440eb0
panic_trap 0
[UNKNOWN:dead_man_sw_handler] 18
[UNKNOWN:timeout_end] 4c
clock 134
i_softmod 2a8
flih_603_patch cc

I do have a dump also if someone knows where to look in it for more
info and/or/cause of dms.

Thanks

2. forwarding netbeui over TCP/IP

3. HACMP problem with serial network link

4. Printing help needed!

5. HACMP -Problem

6. upgrading from 2.2.1 to 2.2.2

7. HACMP problem ...

8. CDRecord problems?

9. HACMP: Problem with adress takeover...

10. HACMP problem

11. Possible Hacmp problem

12. Does hacmp synchronisations need hacmp started ?

13. Difference between HACMP and HACMP/ES