 | netman <netman@ | | NewsGroup User |
| Re: OES NetWare NCS supported ? | 9/18/2008 3:44:04 PM |
 Reply
| 0 |   |
| That's exactly the point !
Robs Site is great and i enjoy his knowledge and informations but i only
can use it in labs or POC's. At least you need the full vendor support
to provide solutions for customers like healthcare, gouverment etc.
Adrian Tritschler schrieb:
> Richard Beels [SysOp] <beels@technologist.com> writes:
>
>> a quick gander at
>> http://pubs.vmware.com/guestnotes/wwhelp/wwhimpl/js/html/wwhelp.htm
>> shows that netware 6.5 is supported on VI3.5U2.
>>
>> Clustering is a core technology in netware and is supported.
>
> Be *very* careful about that. That is what we had assumed from reading
> the literature and talking with assorted vendors.
>
> We have had support calls with both VMware and Novell regarding NetWare
> clusters on VMware ESX and have been told by both parties that while
> NetWare is supported on VMware ESX, *clustering* is not. That is not to
> say that it does not work, just that you may find yourself unsupported
> if something goes wrong.
>
> VMware state that *only* two-node Microsoft clustering is supported, no
> other form of cluster technology.
>
> After pressing our Novell rep. on this point we were provided with a
> document (dated 07Dec2004) stating:
>
> Novell Technical Services (NTS) will provide reasonable best effort
> troubleshooting support for Novell products in VMware
> configurations.
>
> You may find, as we did, that 'reasonable best effort' does not include
> resolving a specific problem, other than with a recommendation to use
> XEN or OES linux!
>
> To include a quote from one of the other followups:
>
>> yeah, i read the Q backwards. actually, the search age not the
>> results page would be better.
>
>> http://developer.novell.com/yessearch/Search.jsp
>
> Reading each of the "VMware ESX" documents returns seems to show that
> they all refer to OES Linux.
>
> If you need support for clustering on NetWare on VMware, make sure you
> get it in writing.
>
> Adrian
>
>> --
>> Cheers!
>> Richard Beels
|
| Adrian Tritschl | | NewsGroup User |
| Re: OES NetWare NCS supported ? | 9/23/2008 2:57:00 AM |
Reply
| 0 | |
| Richard Beels [SysOp] <beels@technologist.com> writes:
> two points...
>
> 1. they're waxing lyrical because it does work.
So we keep being told. I know it works, but I cannot get official
support.
Unfortunately we have a four-node cluster of which two nodes abend
nightly as soon as the backup software puts disk IO under load with "Ate
Poison Pill in SbdWriteNodeTick given by some other node." and which
Novell and VMware will not support.
Even had consultants in on the basis of their experience with NetWare +
clustering on ESX, they confirm that it works, that our system should
work, with the caveat that they've never seen a working cluster of four
nodes or more.
A support call was open for two months with Novell and all we've got to
show for that is advice to use SCSIHD.HAM v3.03.10 with "/RETRIES=6" and
an SBD partition that's been rebuilt four times.
We do have two two-node clusters and a further eight unclustered NetWare
servers under ESX on the same blade farm that have no problems under
normal operations.
Additionally, when the team managing our SAN perform controller resets
on the SAN, then clustered NetWare guests generally abend, sometimes
unclustered NetWare guests lose access to a disk volume and dismount it,
and Windows and Linux guests never have a problem.
> 2. novell is drafting a new public statement; no eta though.
That is interesting, but its likely that we'll now have to be looking at
NetWare or OES Linux under XEN unless the eta is soon, since we've been
told by directors that if the vendor(s) won't state that they'll support
something then we are not to use it in production.
Adrian
> --
> Cheers!
> Richard Beels
|
| Adrian Tritschl | | NewsGroup User |
| Re: OES NetWare NCS supported ? | 9/24/2008 6:26:08 AM |
Reply
| 0 | |
| Richard Beels [SysOp] <beels@technologist.com> writes:
> your experience is not the norm. i've seen 4 node clusters in vmware.
> i know there are larger ones as well. are these all on the same host
> or not?
In our test/dev system there are four HP blades, all running ESX. There
are 10 VMware guest systems, a four-node development cluster, a
four-node test cluster and two additional test servers. Moving the
guests around the hosts has the faults follow the two particular guests
that abend every night.
Looks like the system will be blown away any day now and rebuilt as SLES
with XEN on the hardware and OES linux/OES NetWaer guests.
> i'm with marcel, netware on vmware. linux should be better under zen
> but i haven't used xen much so....
> if high load is casuing poison pills, increase the heartbeat/timout
> settings.
Tolerance, Heartbeat and Master Watchdog are at their defaults(8,1 & 2).
Under instruction from Novell the Slave watchdog was increased from 8 to
16 on July 23, then from 16 to 32 on July 29. To no avail.
How high should we have to increase the settings?
> if you're losing commo to the SAN (just how often do you sAN guys
> reset the SAN?), it's the HBA driver.
The HBA drivers are all bundled in with ESX, there is no option to
upgrade them outside of waiting for an ESX update from VMware. VMware
won't register a defect against the current driver because... they don't
support NetWare clustering.
The SAN guys reset their controllers whenever the IBM SAN engineers tell
them to, or when their maintenance requires them to. Its not that they
do it often, but there are dual controllers, so according to all their
documentation (and experience) resetting an individual controller or
performing LUN rebalancing should be transparent to the end system --
and it is transparent to the end system ... so long as the end system
isn't one of our clustered NetWare boxes.
> if you've had a call open for 2 months with nothing to show, escalate
> the call.
We did, that was when we were told that Novell clustering wasn't
supported on ESX by Novell.
Adrian
--
> Richard Beels
|
|