Dear All,
Lets set the scene first. I have a two node Alpha DS20 cluster running Tru64
Unix v4.0F and TruCluster Available Server v1.6. They both have two network
cards, one network card is connected to the LAN, and the other is only used
as a heartbeat and so is connected to the other node using a cross-over
cable. The network configuration in the cluster is as follows:
Member name: sydcucl1a
Member IP address: 10.1.1.40
Daemon communication use: Primary
Member aliveness ping use: Backup
Monitor interface setting: monitor
Interface name: tom
Interface IP address: 100.1.1.40
Daemon communication use: Backup
Member aliveness ping use: Primary
Monitor interface setting: ignore
Member name: sydcucl1b
Member IP address: 10.1.1.41
Daemon communication use: Primary
Member aliveness ping use: Backup
Monitor interface setting: monitor
Interface name: jerry
Interface IP address: 100.1.1.41
Daemon communication use: Backup
Member aliveness ping use: Primary
Monitor interface setting: ignore
(above taken from asecdb)
ASE Network Configuration
Member Name Interface Name Member Net Monitor
___________ ______________ __________ _______
sydcucl1a sydcucl1a Backup Yes
sydcucl1a tom Primary No
sydcucl1b sydcucl1b Backup Yes
sydcucl1b jerry Primary No
(Taken from asemgr)
So to conclude, the nodes are named sydcucl1a and sydcucl1b, their public
network is 10.1.1.* and their private heartbeat network is 100.1.1.*.
The services are as follows:
10.1.11.20 cpdlv
10.1.11.22 yhsgllv
10.1.11.24 matl
10.1.11.25 wslv
10.1.11.27 nyprinter5
10.1.11.28 cobol
10.1.11.29 cortex
10.1.11.30 prtsvr
prtsvr and nyprinter5 are just used for ip aliases and the rest are disk
services used for oracle databases etc.
Right, now down to my problem. All I want to do is migrate my cluster nodes
and services from the 10.*.*.* (255.0.0.0) to 172.16.2.* (255.255.255.0).
Has anyone got a tried and tested method for doing this? I have tried twice
so far:
* 1st Attempt. Relocated all services to sydcucl1a and removed
sydcucl1b from the cluster. Changed the network address of sydcucl1b
(172.16.2.41) and rebooted. Once back up tried to add back into the cluster
and asemgr would not let me saying that sydcucl1b was in a different subnet.
So, I had to reverse my change and abort mission!
* 2nd Attempt (last night). Relocated all services as above. Made
private network primary and public network backup. Then tried to remove the
backup network so that I could change the network addresses and add the new
back in once I had changed the addresses. The problem was that it wouldn't
let me remove the network address associated with the hostname. Compaq told
me to change the entries in /etc/hosts so that the hostname was associated
with the heartbeat network and visa versa. I then tried to remove the
10.1.1.* network which it let me do. I then tried to save the configuration
and got the message "unable to update the configuration" or something along
those lines. I was then kicked out of asemgr, so I quickly put /etc/hosts
back how it was. I could not get back into asemgr on either node, so I
shutdown sydcucl1b, and again attempted to run asemgr, shortly after which
sydcucl1a crashed. I left sydcucl1b at halt while sydcucl1a booted. Once
booted I was able to get into asemgr and check everything out and all seemed
ok, i.e. back to how it was before, so I brought up sydcucl1b. Everything
was fine, services etc so I decided to again abort my mission.
What is the right way to do it? There must be someone out there who has done
this successfully without having to completely break the cluster and build
from scratch!? It seems to me that the difficulties revolve around the
cluster database.......
Any help will be gratefully received and accepted.
Thanks in advance.
Regards,
Mark
-----------------------------------------------------
Mark C Backhouse
Unix Systems Server Manager
Systems & Network Services
York Health Services NHS Trust
mailto:Mark.Backhouse_at_excha.yhs-tr.northy.nhs.uk
Received on Thu Jul 19 2001 - 11:05:14 NZST