System Management Homepage - Unable to save threshold values

[Brian Reiter]

Hi Folks,

I've been asked to trial SMH and see whether or not that couple with Insight Manager will solve our management requirements. Anyway it all seems to work until the point where we try and set and save one of the threshold values (disk or CPU). At this point the browser interface to SMH just hangs waiting a repsonse back from the host.

The SMH$ADMIN identifier has been granted to the approriate accounts.

The system is:

RX2660 16 Gbyte, 2 CPUs 4 cores

OpenVMS 8-3-1H1 Update 7 (Yes I know)

CSWB 2.2

CSWB_PJP 2.20

SSL 1.4-453

SSL Uodate V1

SMH V2.0-17

SMHPAT V.0-21

I've used SET AUDIT and **bleep**/AUDIT to try and track down problems, but everything claims to be happy.

Hope you can help

cheers

Brian

[Hoff · Accepted solution]

SMH is not my favorite semi-unmaintained system management product.

What you've encountered is fairly typical of my experience with SMH; obscure error messages, instabilities and general all around flakiness.   Check the server logs for any clues.  The setable values are via IPMI via SNMP IIRC, which implies that SMH might not be chatting with SNMP or with your hardware "correctly", or that the SMH$CPQTHRESH process has tipped over. 

Have you followed the instructions around enabling the SNMP "Sets" flag shown by SNMP (see the TCPIP> SHOW CONFIGURATION SNMP/FULL output)  (That's another thing I don't like about SMH.  The configuration sequence is manual and the results are fragile.  If you miss something in the sequence, the resulting errors can be obscure.)  See Appendix B of the installation manual for the SNMP configuration sequence for details.

If it's not the SNMP set-up, which Integrity box is this?  Some of these have limits in what can be monitored.)

If the error is not visible in the logs and if you've verified your SMH configuration sequence, call the HP support center folks and see if they have a troubleshooting sequence available, or if they have some unreleased patches around, or know of something specific to your configuration, or if you've stumbled into a new SMH bug.   

And once you get it operational, see if using SNMP (alone) will resolve your (unspecified) "management requirements", using an SNMP monitoring tool for whatever platform you're basing your remote monitoring tools on.

And FWIW: SMH has been vulnerable to security breaches, and involving the local and remote and escalations flavors, IIRC.  AFAIK, the VMS SMH version still lags behind the Linux and Windows SMH versions, and those platforms have received a number of security fixes.  I've occasionally verified the attacks reported against those other platforms also exist in an VMS SMH implementation that I control, when the exploit code is available.  Given what I've seen of the published vulnerabilities in the security notification lists, I'd not expose VMS SMH to an untrusted network.

[Brian Reiter]

Thanks Hoff,

I went back in and redid the SNMP configuration (perhaps for the third of fourth time), restarted SNMP and SNH and at last things started to work. Some error messages would have been nice, there was nothing in the SMH logs and I didn't check the SNMP logs.

Out of interest is SMH extensible? It seems like it should be, but it would seem that HP are tyring their hardest to make life difficult.

cheers

Brian

[Hoff]

AFAIK, no.

Other than as a path for acquiring SNMP, I don't generally use nor recommend SMH.  Here is why. 

If I needed web-based monitoring, I'd expect that Python and Django or analogous would be far easier, and would likely roll my own using that, and using the Python extensions for OpenVMS and related.  (Django is just amazingly fast for this stuff.)  Or I'd port over something that does what I need.  Alternative solutions here include the (somewhat spotty) Nagios NRPE port for VMS.  Nagios is extensible, and has the option to fix errors and to extend its capabilities, where I don't have that option with SMH.  

Or what I use regularly: SNMP.