Ignite make_net_recovery 2 error: unknow host & swinstall

[Gary L]

Hi

I am currently experiencing some Ignite net_recovery problems, need your helps:
"serverA" is our ignite-ux server, running GUI. I have added over 20 clients into this Ignite server for backup/recovery via "Action" - "Add new client for recovery". But, I met some different failures when I tried to run the "make_net_recovery" for our 4 rp boxes: ServerB, C, D (BTW, others already make_net_recovery completed). Details are as follows:
Erro 1: "serverA is unknown host".
serverB is on the same subnet with serverA. When I run make_net on serverB above erro appear after system "remsh" to B, I keyin the root passwd of B. ServerC on other subnet system inform me "B is not on the same subnet, see the man page for make_boot_tape for information", but the process could be go ahead to the next page seesion, the erro appear after I keyin the root passwd of C. Normally, the process will start up run make_net_r after I keyin the root passwd let server "remsh" to the client.
What's going on and how to handle the "make_boot_tape" I don't wanna touch any serverC's OS re-install etc. serverC is a important production box, I just wanna backup its OS image for failure recovery.

Erro 2: "swinstall failed to install recovery tools from serverA:/var/opt/ignite/deports/revovery_cmds"
ServerD, E are not on the subnet with Ignite serverA. I could add them into Ignite why the system could not swintall the tools.

Any answers will be very very appreciate!

-G

[Ludovic Derlyn · Accepted solution]

hi

for error 1, check your name resolution

/etc/hosts , dns ...

For error 2, modify .rhosts on the client to authorize server ignite

for serverA:/var/opt/ignite/deports/revovery_cmds"
check if product installed on client are identical to server , else remove ignite release

Regards

L-DERLYN

[Gary L]

Additional information input:

I am so sorry about my above erro informations, I made some mistakes, correct as follows:
Erro 1: " serverA is an unknow host"
serverB in the same subnet, the erro msg appear afer I click "OK" at "Allow X windows access by running "xhost +serverB" on .." not keyin root passwd.
serverC not on the same subnet, system inform "not on same subnet and "make_boot_tape", and appear the unknown erro msg after keyin the serverC root passwd.

Erro 2: "swinstall failed to install recovery tools from serverA:/var/opt/ignite/deports/revovery_cmds"
serverD, E not on the same subnet with Ignite A, this erro msg appear at I tried to "Add new client for recovery" after I keyin the client hostname and keyin the root passwd of D and E. I could not add those two clients into Ignite ServerA at all.

/etc/hosts
/etc/nsswitch.conf
/etc/resolv.conf
I have configured for all the HP box and on NFS, I have start server and client, and add entries in /etc/exports and did exportfs -av

[Gary L]

Hi L-derlyn

Thank you very much for your fast reply.
I made some mistakes on erro messages, sorry about it.
For error1, you concern the DNS side. I server aleady did the DNS setting on nsswitch.conf and resolv.conf couple years ago, we could did routine tasks use hostname directly instead of IP. But I met "hostname" issue at the first step of adding clients when I first time added client into ignite-A. So I added all our HPUX box's IP and hostname entries in /etc/hosts. The problems be solved, I could added client. But this "unknow host" issue appear the client be added and I wanna do the "make_net_recovery".
Probably, I did a stupid thing on adding whole IP/host entryies into /etc/hosts. Why Ignite didn't check DNS setting? BTW, my /etc/nsswitch.conf setting was: hosts: files dns nis. Is it fit for Ignite, I am sure it works on other things that related with DNS.

[Gary L]

Hi

I have fixed the first erro. "unknown host"
Yes, it is /etc/hosts problems. But not on server-side. It must be configured on client-side. I add Ignite server A's entry into client server's /etc/hosts, it works.

But I have some recovery concern about serverC that NOT on the same subnet with Ignite-serverA. I have created serverC's archive on serverA. But Ignite doc write: I should use "make_boot_tape" create a tape for the other subnet. But we don't have tape drive at all. We only have two MSL6030 tape libraries for Data backup via Hp Data Protector. Is there any problems on recovery the span subnet serverC?

who could tell me how to fix erro2? Add new client problem?

thank you very much

-G

[Gary L]

Hi

Probably, I know the reason of error 2.
serverC and D could not be added into ignite serverA, caused by the Ignite-ux version of C and D higher than the serverA. C and D version is C.6.1.44, and serverA is B.5.4.50.

[Bill Hassell]

swinstall and ignite are critically dependent on proper network setup in each machine. It sounds as if your DNS server is not properly setup at all. But that is easily fixed by adding *all* of your clients and servers into /etc/hosts and copying this /etc/hosts file to every system. Then make sure your nsswitch.conf file has this line in it:

hosts: files [NOTFOUND=continue UNAVAIL=continue] dns

This means that all hostname lookup starts with /etc/hosts which is under your control.

AS far as ignite and swinstall paying attention to the host and nsswitch file, that is never done. Instead, like all network programs, they use system calls to gethostbyname() and gethostbyaddr() and IP address resolver in HP-UX uses these files to provide IP addresses.

[Gary L]

Hi Bill

How are you doing. I got lots of help from you, thank you very much.

I have set nsswitch.conf long time ago, (hosts: files dns nis). But ignite didn't go ahead to search from DNS when it wasn't found the host in /etc/hosts. whether I must write (files [NOTFOUND=continue UNAVAIL=continue] dns)? whether I could write it like this: hosts: dns files, let the system check DNS server first and check the file later if DNS server down.

I have a important question that asked twice above, would you please do me a favor:
serverC and D on another subnet, although I could created and saved their archives in serverA, but whether those two client recovery like others, in other words, if severC D crash, whether I could do: cold startup C, go into main menu, bo lan.n.n.n.n install boot from Ignite serverA and find the img archive to recovery serverC? Ignite said using make_boot_tape, but we don't have more tape drive.

How to backup serverA, through client Ignite-ux as server, add client A and do the img backup? I tried, but met NFS mount issue erro: could not NFS mount dir /var/opt/igite?client ... Failed omounting dir /var/opt/ignite/recvery/client_mnt/co refuse errno=239, failed to creat NFS mount dir for config file.
If serverA already be added into Ignite ServerA and done the img backup. Now, I wanna backup A on another server, is it the above erro's reason? How to fix the NFS issue.

So many question, thanks again

-G

[Bill Hassell]

> But ignite didn't go ahead to search from DNS when it wasn't found the host in /etc/hosts.

As I mentioned, Ignite never searches hosts or DNS. This is done by the resolver inside HP-UX. Why is this hostname/IP not in /etc/hosts? I would not rely on DNS for production machines, especially in a recovery or reinstall situation.

> whether I must write (files [NOTFOUND=continue UNAVAIL=continue] dns)?

Yes. Change you nsswitch.conf file.

> whether I could write it like this: hosts: dns files, let the system check DNS server first and check the file later if DNS server down.

WORST possible choice. When you go to DNS first, you are relying on someone else to configure (and keep configured) critical addresses, and as you have already seem, DNS is failing. Note also that if a DNS server listed in your HP-UX box goes down, there will terrible delays for everything that uses DNS, typically 20-30 seconds for every down server request (and there may hundreds every minute).

If you want to test your DNS server, use these 4 commands where mycpu has the address 1.2.3.4

nslookup mycpu
nslookup 1.2.3.4
nsquery hosts mycpu
nsquery hosts 1.2.3.4

ALL of them should return a correct result before you try to get Ignite working.

> ...cold startup C, go into main menu, bo lan.n.n.n.n install boot from Ignite serverA

If C is located on another subnet, then no, you cannot see server A at all. You need a small kernel loaded from a local Ignite helper machine.

> ...could not NFS mount dir...

You have NFS problems that must be resolved, but do not attempt to fix NFS until your resolver is working (hosts and DNS).

[Gary L]

Thanks Bill.

if the servers that located on other subnet with Ignite-Sever couldnot be recovery, even couldnot be seen at all when the client reboot and do the "search lan install", then what did you do normally in my situation? Using D backup C via Ignite serverD via make_net_recovery? (we don't have tape drive any more)

thanks again.
Happy weekend

-G