SEO

Hi,

I am experimenting with Sun Management Center 3.0 as I am under the
impression that it allows me to monitor hardware components on my sunfire
machine.
However, I dont seem to be able to find the MIBs for doing so. is anyone
aware if they
are still available for download.

regards, Veni

Hi Veni,

> I am experimenting with Sun Management Center 3.0 as I am under the
> impression that it allows me to monitor hardware components on my sunfire
> machine.
> However, I dont seem to be able to find the MIBs for doing so. is anyone
> aware if they are still available for download.

First off, any reason why you're not using a much newer version of
SunMC? Stuck with the version supported by SRS? SunMC's up to 3.5.1
now. Also, all this information is viewable through the SunMC Console:
are you trying to poll this info from another framework? There are
already integrations available for HPOV/CA/Netcool/Tivoli...

Sun publishes several MIBs for SunMC, although none are for the
hardware:

http://wwws.sun.com/software/solaris.../snmp.html#0q2

....although there are posts in the SunMC forum of people
reverse-engineering the MIBS:

http://forum.sun.com/thread.jsp?forum=181&thread=15248

....or if you're looking for something off-the-shelf:

http://www.halcyoninc.com/products/Frameworks/index.php

Regards.

Note: I am an employee of Halcyon (www.HalcyonInc.com)

Thanks Mike for answering. I am simply using the freeware SMC 3.0 because it was the only copy I could source for. downloading it seems impractical given the size. We do not intend to implement SRS because its an enclosed environment. From the documents from sun, it is implied that from Symon 1.6 http://docs.sun.com/db/doc/805-4829/6j4io2pah?a=view, we should have the mibs for hardware component monitoring, but I havent really found what I want from the sun site or http://www.mibdepot.com. after all the hassle, I am wondering if it would simply be cleaner to implement an snmp trap (for hardware component monitoring) using the generic tools on sun & perl given that critical errors and failures should be readable and grepped off some of the syslog or var/admin logs.


regards, Veni
inveni@hotmail.com




"Mike Kirk" <primealert@gmail.com> wrote in message news:f0bc53ee.0410121254.48f34b15@posting.google.c om...
> Hi Veni,
>
> > I am experimenting with Sun Management Center 3.0 as I am under the
> > impression that it allows me to monitor hardware components on my sunfire
> > machine.
> > However, I dont seem to be able to find the MIBs for doing so. is anyone
> > aware if they are still available for download.
>
> First off, any reason why you're not using a much newer version of
> SunMC? Stuck with the version supported by SRS? SunMC's up to 3.5.1
> now. Also, all this information is viewable through the SunMC Console:
> are you trying to poll this info from another framework? There are
> already integrations available for HPOV/CA/Netcool/Tivoli...
>
> Sun publishes several MIBs for SunMC, although none are for the
> hardware:
>
> http://wwws.sun.com/software/solaris.../snmp.html#0q2
>
> ...although there are posts in the SunMC forum of people
> reverse-engineering the MIBS:
>
> http://forum.sun.com/thread.jsp?forum=181&thread=15248
>
> ...or if you're looking for something off-the-shelf:
>
> http://www.halcyoninc.com/products/Frameworks/index.php
>
> Regards.
>
> Note: I am an employee of Halcyon (www.HalcyonInc.com)

Veni,

> From the documents from sun, it is implied that from Symon 1.6
> http://docs.sun.com/db/doc/805-4829/6j4io2pah?a=view, we should have
> the mibs for hardware component monitoring

Ah, I see where you're coming from. SyMON 1.x and SyMON 2.x (renamed
to "Sun Management Center" as of version 3) are totally seperate
products. Essentially 1.x was scrapped when 2.x was codeveloped with a
partner (where Halcyon comes in). 2.x onwards all use the same core
infrastructure, so 2.x docs still have some relevance to the most
current version (3.5.1). But 1.x docs have nothing to do with 2.x or
higher.

> I am wondering if it would simply be cleaner to implement an
> snmp trap (for hardware component monitoring) using the generic tools on
> sun & perl given that critical errors and failures should be readable
> and grepped off some of the syslog or var/admin logs.

Scripts and logs may well be enough for your environment, especially
if you have a bunch of non-production and/or small systems to
maintain. But SunMC hardware monitoring also uses C code/libraries to
query hardware for things that otherwise don't show up in log files.
Especially for larger systems (3/4/6800, 12/15/20/25k range).

You also have to ask yourself how much work it will be to maintain
your setup in the face of new Sun hardware and software (i.e. Solaris
10) given that the basic hardware monitoring in SunMC is free.
Downloading a new version may be easier to do than hacking away at
Perl, though not nearly as fun!

Regards,

Mike

Note: I am an employee of Halcyon (www.HalcyonInc.com)

That makes perfect sense now! The documentation in symon 1.6 doesnt tally
with SMC3.0.
I've been trying to learn how to work out my own implementation, but there
arent any books or references to troubleshooting or writing your own snmp
agents. mauro's essential snmp & Stallings' SNMP, SNMPv2,v3, RMON1&2 have
been useful in deriving a coherent overview.. but you really dont get to
know enough to troubleshoot mibs or snmp problems. Hopefully, I may get a
better idea how to write my own mib after going through "understanding snmp
mibs". But without smc or Symon, how would you grep for critical events then
if all not such events are logged?

regards, Veni

inveni@hotmail.com


"Mike Kirk" <primealert@gmail.com> wrote in message
news:f0bc53ee.0410130727.3d59407e@posting.google.c om...
> Veni,
>
> > From the documents from sun, it is implied that from Symon 1.6
> > http://docs.sun.com/db/doc/805-4829/6j4io2pah?a=view, we should have
> > the mibs for hardware component monitoring
>
> Ah, I see where you're coming from. SyMON 1.x and SyMON 2.x (renamed
> to "Sun Management Center" as of version 3) are totally seperate
> products. Essentially 1.x was scrapped when 2.x was codeveloped with a
> partner (where Halcyon comes in). 2.x onwards all use the same core
> infrastructure, so 2.x docs still have some relevance to the most
> current version (3.5.1). But 1.x docs have nothing to do with 2.x or
> higher.
>
> > I am wondering if it would simply be cleaner to implement an
> > snmp trap (for hardware component monitoring) using the generic tools on
> > sun & perl given that critical errors and failures should be readable
> > and grepped off some of the syslog or var/admin logs.
>
> Scripts and logs may well be enough for your environment, especially
> if you have a bunch of non-production and/or small systems to
> maintain. But SunMC hardware monitoring also uses C code/libraries to
> query hardware for things that otherwise don't show up in log files.
> Especially for larger systems (3/4/6800, 12/15/20/25k range).
>
> You also have to ask yourself how much work it will be to maintain
> your setup in the face of new Sun hardware and software (i.e. Solaris
> 10) given that the basic hardware monitoring in SunMC is free.
> Downloading a new version may be easier to do than hacking away at
> Perl, though not nearly as fun!
>
> Regards,
>
> Mike
>
> Note: I am an employee of Halcyon (www.HalcyonInc.com)

Veni,

> But without smc or Symon, how would you grep for critical events then
> if all not such events are logged?

I think it ends up being that Sun and it's engineers are always going
to know more about their hardware than anybody else, and the ways they
query the status of that hardware aren't neccessarily going to be made
public. But you're approach is still going to get you quite a bit of
the info you want.

SunMC gives you all this info, in software that is free, and is
supported for all new Sun gear and Sun OS's. So I don't understand why
you wouldn't use it? Or is this more of a learn-how-SNMP-works
project?

Also, have you looked at Solaris 10 yet? From what I've heard it will
have some sort of "predictive self-healing" daemon on every machine...
so maybe the events that it actions on will be wide open for you to
read? That may make things easier.

Regards,

Mike

Its more because I am using HP Openview and I'd like to be able to either
poll for the hardware component status (that most of my components are still
running, even the redundant power supplies). Thus, I had assumed (wrongly)
that the hassle free method was to install SMC which would have its own snmp
agents and then pass the mib to HpOpenview.
I dont think my peers would agree to an OS change or upgrade. Thus, I am
limited to superficial amendments. Given that the status of power supplies
and scsi disks arent necessarily found in syslogs or /var/admin, I suppose
another way may be to run ./explorer and the grep for the values in my
explorer output using perl. Not a very intelligent way to go about it
though.

Anyway after reading the documentation on SMC3.0, it seems to me that there
may not be an actual "hardware" mib that I can parse to my NMS.

Ps. Does anyone know of HPopenview compatibility issues with solaris? Not
all my mibs work, but I'd like to think that the mibs I am using conform
pretty much to the RFCs.

regards, Veni
inveni@hotmail.com

Hi Veni,

> Its more because I am using HP Openview and I'd like to be able to either
> poll for the hardware component status (that most of my components are still
> running, even the redundant power supplies).

Now I understand, wish I knew you had OpenView earlier: lots of people
push Sun hardware status to HPOV, but polling isn't the way to do it.
SunMC is built specifically to integrate with other frameworks like
Tivoli/CA/HPOV/Patrol/Netcool etc. It will actually "push" (by an SNMP
trap or Java message) any hardware alarms to you.

The push model is way more efficient, because you aren't continuously
polling dozens of data points from all your boxes... thus saving
network bandwidth and cpu. And you don't have to load a bunch of
random OIDs into NNM

> I dont think my peers would agree to an OS change or upgrade.

I'm guessing you're talking about changing from HPOV to SunMC, not
from HP-UX to Solaris, and you're exactly right: after spending big
$$$ on HPOV nobody in their right mind would rip it all out. What most
places do is start off by installing the free SunMC Agent on every
box, in parallel to HPOV (each Agent is about 10-15MB of memory, and
less than 1% cpu, they're tiny) and use it only for hardware. You
still leave HPOV everywhere, because it's probably still watching you
apps. It's just that SunMC is much better for Sun hardware.

Then, once you're happy it's all up and running (go yank a power
supply out: make sure HPOV gets the message), you may choose to use
SunMC for more (i.e. it has better Solaris kernel monitoring than HPOV
as well). Or not. It's up to you.
Because HPOV is just using SunMC as a data collector, you also have a
bit of flexibility: i.e. perhaps if you run VCS the monitoring for it
may be better or cheaper on SunMC then HPOV... you can buy the version
you want.. because it all ends up in HPOV anyways. Make your Sun and
HP software reps fight to give you a better deal

Anyways, back to what you need: either look into the HP-provided
integration:

http://support.openview.hp.com/cpe/p...mc.jsp?print=1

....or Halcyon-provided:

http://www.halcyoninc.com/products/HPOpenview/index.php

I'm partial to the Halcyon version, because I work for them But
everything I've said it my own opinion blah blah blah....

Hope this points you in the right direction.

Regards,

Mike

I am kinda stuck with HPOV. Thus, I have to figure out a way to get my machine to be monitored by HPOV. And I agree with you, after reading through the documentation, SMC seems more robust. For now, I have to either come up with my own MIB or source for a way where HPOV gets a trap whenever theres a critical hardware failure. Ps, since we cant grep for all the values via var/adm and syslog. would prtdiag or running the explorer be the best way to check for hardware status?

regards, Veni
inveni@hotmail.com


"Mike Kirk" <primealert@gmail.com> wrote in message news:f0bc53ee.0410160430.332695b2@posting.google.c om...
> Hi Veni,
>
> > Its more because I am using HP Openview and I'd like to be able to either
> > poll for the hardware component status (that most of my components are still
> > running, even the redundant power supplies).
>
> Now I understand, wish I knew you had OpenView earlier: lots of people
> push Sun hardware status to HPOV, but polling isn't the way to do it.
> SunMC is built specifically to integrate with other frameworks like
> Tivoli/CA/HPOV/Patrol/Netcool etc. It will actually "push" (by an SNMP
> trap or Java message) any hardware alarms to you.
>
> The push model is way more efficient, because you aren't continuously
> polling dozens of data points from all your boxes... thus saving
> network bandwidth and cpu. And you don't have to load a bunch of
> random OIDs into NNM
>
> > I dont think my peers would agree to an OS change or upgrade.
>
> I'm guessing you're talking about changing from HPOV to SunMC, not
> from HP-UX to Solaris, and you're exactly right: after spending big
> $$$ on HPOV nobody in their right mind would rip it all out. What most
> places do is start off by installing the free SunMC Agent on every
> box, in parallel to HPOV (each Agent is about 10-15MB of memory, and
> less than 1% cpu, they're tiny) and use it only for hardware. You
> still leave HPOV everywhere, because it's probably still watching you
> apps. It's just that SunMC is much better for Sun hardware.
>
> Then, once you're happy it's all up and running (go yank a power
> supply out: make sure HPOV gets the message), you may choose to use
> SunMC for more (i.e. it has better Solaris kernel monitoring than HPOV
> as well). Or not. It's up to you.
> Because HPOV is just using SunMC as a data collector, you also have a
> bit of flexibility: i.e. perhaps if you run VCS the monitoring for it
> may be better or cheaper on SunMC then HPOV... you can buy the version
> you want.. because it all ends up in HPOV anyways. Make your Sun and
> HP software reps fight to give you a better deal
>
> Anyways, back to what you need: either look into the HP-provided
> integration:
>
> http://support.openview.hp.com/cpe/p...mc.jsp?print=1
>
> ...or Halcyon-provided:
>
> http://www.halcyoninc.com/products/HPOpenview/index.php
>
> I'm partial to the Halcyon version, because I work for them But
> everything I've said it my own opinion blah blah blah....
>
> Hope this points you in the right direction.
>
> Regards,
>
> Mike

Hi Veni,

> I am kinda stuck with HPOV. Thus, I have to figure out a way to get my
> machine to be monitored by HPOV. And I agree with you, after reading
> through the documentation, SMC seems more robust.

A couple of posts back you said you were willing to install SunMC if
you could poll it's MIB from HPOV. But now that I've told you that HP
actually builds a free SunMC <-> HPOV integration package, you don't
want to install SunMC at all?

> For now, I have to either come up with my own MIB or source for
> a way where HPOV gets a trap whenever theres a critical hardware failure

The second link I sent you was to have SunMC send HPOV a trap whenever
theres a critical hardware failure.

> Ps, since we cant grep
> for all the values via var/adm and syslog. would prtdiag or running the
> explorer be the best way to check for hardware status?

That would probably be better than nothing. But still, every time you
get a different Sun box, or upgrade Solaris, you'd have to edit the
scripts you made to do all that extra parsing.

a) SunMC gives you the best hardware monitoring of Sun gear (and lots
more)... for free.
b) HP provides a SunMC integration package to get all that good info
into HPOV... for free
c) By using these free vendor supplied tools, you do not have to waste
you time putting together a bunch of scripts that at best would only
approximate their functionality, and you don't have to waste your time
again upgrading those scripts every time Sun or HP changes their
hardware or software

Yes, I work for Halcyon (although this post is my own opinion and
doesn't represent my company), so you can paint me as a pro-SunMC
person, but to me it seems an easy decision to make. Unless your job
description is "to get paid to poorly reimplement the functionality of
other free sofware packages".

Hope this finds you well,

Mike