Insight Management agent do not start

[EricPH]

Hello,

during upgrade from Agent 3.5 to 3.7 (or 3.6 too) on several ES47, I have trouble on 2 of them.
hpsmhd doesn't seem to start, as I am running :
setld -l cpqim370 (same as /sbin/init.d/hpsmhd start)
The process is stuck on :
hpsmd: Could not determine the server's fully qualified domain name, using XX.XX.XX.XX for ServerName
Connected to 127.0.0.1:2301
and is waiting indefinetly...
Strange behaviour, because on other identical computer (hardware and OS) the upgrade went well.

Any suggestion will be greatly appreciated...

Thanks

[Ivan Ferreira]

>>> hpsmd: Could not determine the server's fully qualified domain name,

Check your /etc/hosts file, you should have one entry with the hostname + domain for your host, for example:

192.168.0.1 thihost.domain.com thishost

Also check if the hostname command returns the hostname + domain (fully qualified domain name).

[EricPH]

Thanks for your quick answer.
I have to precise that my computers don't have FQDN, they do not use DNS name. So, in /etc/hosts, I only have :
xx.xx.xx.01 hostH
xx.xx.xx.02 hostS
xx.xx.xx.03 hostG
and all /etc/hosts files are identical.

What is strange, is that installation fails on 2 of 3 computers, the 3rd one (hostH) has the same OS installation (disk clone). I also have the warning on it (hpsmd: Could not determine the server's fully qualified domain name, using XX.XX.XX.XX for ServerName
) but the installation continues to the end.

hostH and hostS are strictly identical (ES47) with 2 network cards on 2 separated networks (1 main, the other ties both computers directly), hostG is a dual ES47 (2 drawers) with 2 network interfaces in an agregate on main network.

Yes, I thinks that the pb is around the fact that hpsmhd launch itself on a bad NIC...
How to check that ? where is the file specifying which interface to use ? (if it is the good diagnostic...)

[Pieter 't Hart]

you could modify the hosts file as :
192.168.0.1 thishost.local thishost

[Pieter 't Hart]

You say you have two networks
name.
>>> hostH and hostS are strictly identical >>> (ES47) with 2 network cards on 2
>>> separated networks (1 main, the other >>> ties both computers directly)

but this does not reflect in the hostfile you supply!
>>> xx.xx.xx.01 hostH
>>> xx.xx.xx.02 hostS
>>> xx.xx.xx.03 hostG

I would expect something like :
xx.xx.xx.01 hostH-nic1-net1
xx.xx.yy.01 hostH-nic2-net2
xx.xx.xx.02 hostS-nic1-net1
xx.xx.yy.02 hostS-nic2-net2

[EricPH]

yes, you are right about the hosts file. But, for me, I was not considering the 2nd network since it is limited to this 2 computers with this 2 NIC (direct connections with a cross cable).
Those netwwork I'm interrested in, is the "1st" one, because it is there that I have my Insight Manager Server (another computer under Windows 2003 srv) which is supposed to collect events...

[Pieter 't Hart]

you could check the /etc/rc.config if NETDEV_0 and NETDEV_1 are mapped to the netdevices same order
something like netdev_0 to ee0 for the main link and netdev_1 to ee1 for the crosslink.

[EricPH]

Thanks for your suggestion.
I'm not sure about this solution, because on the computer without trouble (hostH), I've :
> rcmgr mget | grep -i netdev
...
NETDEV_5=
NETDEV_4=
NETDEV_3=
NETDEV_2=
NETDEV_0="bcm0"
NETDEV_1="bcm4"

as on hostS.

and on the hostG (with nic combined by lag), this variables or not defined :

> rcmgr mget | grep -i netdev
...
NETDEV_5=
NETDEV_4=
NETDEV_3=
NETDEV_2=
-nil-

It will be the end of day work, here, but if you've got another good ideas, please suggest :-)

[EricPH]

Some news...

It has "fall" on start on the second host without the lag on the 2 nic...

but on this hostG, no way...

[Pieter 't Hart · Accepted solution]

I've no experience with lag, but I do expect NETDEV_0 and NETDEV_1 in the rc.config file.
I do have configured netrain in wich NETDEV_0 is associated to nr0 :

NR_DEVICES="1"
export NR_DEVICES
NRDEV_0="nr0"
export NRDEV_0
NRCONFIG_0="ee0,ee1"
export NRCONFIG_0
NUM_NETCONFIG="1"
export NUM_NETCONFIG
NETDEV_0="nr0"
export NETDEV_0
IFCONFIG_0="10.1.3.26 netmask 255.255.0.0"
export IFCONFIG_0

For use with lag I would expect similar entries in this file.
as in http://unix.derkeiler.com/Mailing-Lists/Tru64-UNIX-Managers/2004-01/0147.html
i see conformed that one interface is configured through rc.config and the lag is formed by adding interfaces through /etc/inet.local