check hardware servers

[pascal]

Are there also tests to check the hardware of servers, diskdrives, tapedrives ......

[Guest]

There are no common built-in methods in HostMonitor. I think everything depends on developer of the server and hardware. If they provide some standard interface (e.g. SNMP), its good. If not...

Regards
Alex

[SABIC-EPC]

Alex,

We provide some standard interface (e.g. SNMP), now we use also Compaq Insight Manager, this works also with SNMP traps.
But I don't want to use it anymore ( CIM )
I want to use Hostmonitor to check hardware if it's posible.

Regards,

Pascal

[hmo]

Hostmonitor is not hardware specific in any way - ie. you can use what ever you like, to monitor snmp compatible units. In Hostmonitor you can drag snmp values and have a very secure way of monitoring any hardware device you want. Like Compaq's insigt manager IBM also has it's own - Netfinity etc. If Hostmonitor should contain all this GUI-work it would (imho) loose all it advantages. Monitoring should be quick, simple and very secure. Try to look-up any snmp-enabling for your servers (hardware). You'll be supprised of what you can see!
Regards,
Hans Mosegaard

[Marcus]

[hmo]

If we're talking about Compaq hardware - only - I would agree. In our datacenter we have multible vendors, so it would we a nightmare to work with different specific "insigt managers" etc. I dont know if you only has Compaq-units... but if you dont, then a 3'rd. part SNMP-trapper would also do the job... from here You can - in Hostmonitor - collect all kinda data you want.
Cheers,
Hans Mosegaard

[Marcus]

We have IBM, Dell and Compaq / HP and use the vendor's monitoring tools (besides Openview). It just has a lot of advantages to use the vendor tools. I can't imagine you will create your own snmp jobs to collect all data from the servers.

But then again, that's why I wrote you must have another mechanism to collect the traps

[gdvl]

Hi Marcus en co,

Can you tell me how you use / position CIM, OpenView and Hostmonitor ?

My boss want to use a big one as OpenView or MOM or Argent Guardian or ... . He will have a central collection of ALL events, alerts, history, ... in one big database.
Now we have Hostmonitor, and I'm very satisfied with it as monitoring and alerting tool. Hostmonitor is much much better than the monitoring capabilities of the 'big' ones.

Can you tell me how you use these different parts together ? And what's your experience with them ? (especially conceptual, not only technical)

PS: I think we better mail about this.

Regards,
Gert De Vleeschouwer
mailto:gdvl@concentra.be

[Marcus]

SNMP traps are sent to two locations:
Vendor Hardware tool + Openview

Vendor tools wil forward all events to Openview, this wil result in duplicate messages in openview (which is no problem if you collect the duplicates).

Hostmonitor wil send an snmp trap on every bad status found to the openview server. It wil send a one time only snmp trap to openview when the status changes back to normal.

This results in a single monitoring interface: Openview (we use the javagui).

And the tools from the 'big-ones' are not less, they are not default included

[hmo]

[Marcus]

-Well, knowing the serials is fine when hardware support is needed. Knowing the exact model is even finer. And knowing this when the remote server is offline is a blessing
-Replacing a cpu because it is slowly failing, is nice.
-Replacing a disk when it fails from your array is nice (you don't need an array if you can't see the problem).
-Timing problems or failing memory cards is nice to see.
-A power fan not running, is nice to see.
-Tempatures rising inside the server, is nice to see.

All above did actually happen. If I had to create my own snmp queries, I would have missed them for sure.

Most traps wil reach your management server (if you have one), but when using the vendor tools, all traps are translated to human understandable messages. Looking up or implementing your own (imported) mib translations is a hell of a job. Since you don't actually need a full blown server to implement the vendor tools, costs should not be the bottleneck......

[hmo]

Right on Marcus. We're talking about vendor specific information piped into hostmonitor- or not? Sure I like to know when "anything" is wrong on any of our servers, but I like to approach this by KISS method - ie. do the basic test "you can" with Hostmonitor. On top of that - do run any vendor specific software you like locally on a given server. We primary use IBM and Fujitsu-Siemens servers, and all of them will flash a yellow or red light on the front when anything goes wrong. In that case... we do inspect our servers a couple of times each day... if needed we'll bring up the current vendor-tool to have a closer look - and like you - solve the problem based on "human error-codes". What I dont like is to build up a huge christmas-tree based on many different systems - and link those together in a scarry way (who to trust?).

What I'm talking about is to take a pragmatic approach into the easy lane. I wont say that "your way" is wrong...neither is my way... we just gain the same result in different ways!

Ohh... I would'nt miss my day-walk through our zones... just the sight of "that many" servers flashing (mostly green) is a dream-tour!

Cheers,
Hans Mosegaard

[Marcus]