nworks 5.5 RC1

[krowczynski]

Hi everybody,

would like to know your opinion if I should install the RC1 in my production enviroment, or better wait until the final release will be available?

Thx for your opinions!!!

[arseny]

Hi there,

My opinion is: "Please go ahead with RC1 and enjoy the new functionality!" Only concern I'll have to mention here is support: although we will be supporting your production as an exception if you decide to leave RC there, if there is anything different in GA than in RC from code/issues perspective, we will enforce the upgrade in some reasonable amount of months after GA.

Don't forget, support@veeam.com and your Sales Representative are here to help you out with nworks installation, - so always include Veeam representatives in the project loop.

Cheers,

[krowczynski]

Hi,

thank you for infirmation ynd your opinion.
I think the next day I will try the new RC.

I will let you know how I fell with it.

[arseny]

Hi Arkadiusz,

Sure thing, go ahead and post your impressions in this thread. Don't forget you can also always open support@veeam.com ticket, that would have dramatically lower response time than forum communication.

Have a good day,

[krowczynski]

Hi,

the sys requirement says that the VEM Server must be an Windows 2008.
Now our Server where all components are installed togother is an 2003 x64 server.

So should I install here a brandnew server with 2008 R2 Operating System?

[Irina Simkina]

Hi Arkadiusz,

the system requirements for VEM Server are the following:
Architecture: x86, x64
OS: Microsoft Windows Server 2003 (SP2 recommended) OR Microsoft Windows Server 2008

So you can use your existing environment with 2003 x64 server installed - It's fully supported.

Irina

[krowczynski]

Hi,

thank you for this information.

[krowczynski]

HI,

just updated my enviroment with 5.5 RC1.
Update worked perfectly without any issues.

I will let know everybody more if I get more data in the next days.

[Alec King]

Hi Arkadiusz,

That's great, thanks for the update! Glad you deployed the 5.5 RC successfully

Please let us know how it goes.

Thanks again!

Regards
Alec

[krowczynski]

Hi,

have a few questions, mybe someone can explain it for me.

At first the release go fine, until now no problems, but now here ae my few quesstions.

In the last Version (5.03) I got sometimes messages (at most during a backupphase of a vm) "VM XXX is using 88% allocated cpu ressources.

Since the the release I didn't get this message, so my question is is it a falilure or get the nworks better?

I've also noticed that now the cpu ans memory messages were collected if the are going into a red state and back from vcenter.
I've disabled this messages because a vom often get into a red state and after back into thr normal one.

Also would like to know explained the new message

"VMGuest 'XXX' (on YYYYY) has exceeded threshold for total disk IOPS over 2 samples by using 232 IOPS"


Thank you very much for information.

[Alec King]

Hi Arkadiusz,

For your first question - I believe nworks just got better

The performance Monitors are now based on consecutive or averaged samples- this means that we do not alert on a single cpu usage spike. Out-of-box we will alert if the metric is over threshold for TWO samples - and you can override thу number of samples, using SCOM overrides in the usual way.

Agree with you on disabling the CPU & Memory Alarm events from vCenter - our performance Monitors are much more flexible anyway. (When we ship 5.5 those VC Alarms will be disabled out-of-box too)

For diskIOPS - this is a new metric, and new Monitor, in 5.5. I would really like your feedback!
We take disk reads + writes and calculate total input+output per second.
We're still researching the final threshold for 5.5 release - it's not easy to set a threshold for everyone, because workloads inside a VM can vary, and also performance of disk subsystems varies too.
Are you getting many alerts from this? What kind of values do you have for diskIOPS for your Vms? You can see this in the new Disk I/O VM Dashboard view.

Thanks for the feedback!
Cheers
Alec

[krowczynski]

Hi,
thanks for you fast reply and your answers.


Yes I get a lot of messages because of IOPS, I will send you some values in the next day I hope, because I deleted all my mails because of it!!

[krowczynski]

Hi Alec,

so lets take as sample the mails I got during the night.
Our enviroment is an iSCSI one, with HP SAS Storage. (Just as information for you)

I got aroung 15 Mails and the IOPS were alway above 100.

Once Server was 186 another was 285, it it different and I also think that it depends if at this time the vm was backup up maybe.

Just this as information for you.

[Alec King]

Hi Arkadiusz,

Thanks for the info!
I think you are right, overnight there could be some jobs (such as backup, update, virus-scan etc) that cause high disk IOPS.

285 IOPS is pretty high, I think - that could require investigation. What application runs inside that VM? However there are many factors, and you can decide that it is normal for your environment (especially if it does not stay around 285 all the time)

If you look at the VM Disk I/O Dashboard (in VMware Virtual Machine\ Performance Views) - how long do these VMs stay above 100 IOPS? How many samples? (5 mins by default).

If you look at the Host Disk I/O Dashboard (in VMware ESX Host Server\ Performance Views) - what is the totalDiskIOPS at the host level during these times?
If you drill-down into \disk performance views you can also see diskIOPS for each Host vmhba.

You might find after investigation that you can re-schedule VM backup jobs, virus-scans or patching/updates so that VMs on same Host, or same storage, do not all demand high disk I/O at the same time.
This is one purpose of the nworks MP - to help you optimise your environment

But anyway, you also have the option of using SCOM Override(s) if you decide that the high IOPS values are normal operation for you. You can -
1. increase the default threshold of 100
2. increase the default sample count of 2

Please let me know how it goes - this is great feedback for me!
Thanks again!

[krowczynski]

Hi Alec,

can I also post you e.g. two screenshots?

[Alec King]

Hi Arkadiusz,

Yes please! Screenshots would be great.

Not sure that screenshots/attachments are supported in this forum - so we can just continue in email if that's ok.

I'm -
alec dot king at veeam dot com

Thanks!

[krowczynski]

Hi Alec,

you got mail, I hope!

[krowczynski]

Hi Alec,

any news when nworks 5.5 will be officially released?

[Alec King]

Hi Arkadiusz,

It's coming very soon!

We actually posted a Release Candidate #2 build on veeam.com yesterday - this really is the final build, I believe.

I would watch your inbox, and veeam.com, for announcements over the next couple of weeks....

Cheers,
Alec

[krowczynski]

Hi Alec,

can I download the RC2 already?
I will watch my postbox the next days.!!

[Alec King]

Hi Arkadiusz,

Yes you can!
It's here - http://www.veeam.com/go/nworks-microsoft-vmware-55rc2

Note - you will have to delete the RC1 MP from SCOM, before you can import the RC2 MP. Upgrade RC1 to RC2 is not supported (however, we plan for upgrade from RC2 to 5.5 release version will be supported)
You should also upgrade all other nworks components to RC2 level (Collector, Enterprise Manager, and UI)

Let me know if you have any questions on RC2 - and keep watching your postbox!

Thanks,
Alec

[krowczynski]

Hi Alec,

I have a question.
I've created a second Clustern in my vcenter, everything works fine.
In the nworks Enterprise Manager I can see the 2nd Cluster, but wehen I look in my SCOm under nworks Cluster or other points I cannot see the new created cluster and also not he newly installed ESX hosts.

Help would be glad!
Thanky ou in advance.

[Alec King]

Hi Arkadiusz,

Well, certainly the discovery of new Cluster should be automatic....

In the nworks UI, is the Cluster (and all Hosts) checked? (in Connected Servers tab).
And is the Cluster seen under a Collector as a Monitoring Job? (in Enterprise Manager tab)

Do you see any errors or warnings in the nworks view ''Enterprise Manager Dashboard'' in SCOM?

You can try the Rebuild Full Topology button in nworks UI (seen when you select Enterprise Manager node in the Enterprise Manager tab).
After several cycles (maybe 15 - 20 minutes) then all Clusters, Hosts and VMs should be rebuilt in SCOM.

If that doesn't work - let me know,
Thanks!
Alec

[krowczynski]

Hi Alec,

thank you for answer.
After doing a rebuild and a reboot of scom and nworks everything is fine.
By the way, when will be 5.5 oficially released?

[Alec King]

Hi Arkadiusz,

Good, everything is working again!

And for nworks 5.5 release....I can't promise exact dates.....
But I am pretty sure you will see something this month - Keep watching your inbox!

Thanks!
Alec

[krowczynski]

Hi,

Alec thank you for this information, because I see there are a few bugs in RC1.
I see some warning and after that if I look in my vsphere Client everything looks okay, but it is RC1, so I think it is so.

[krowczynski]

Hello Alec,

in which way I have to upgrade from 5.5 RC1 to 5.5.1?

Thank you for some information.

[vBPav]

Hello Arkadiusz,

I will have to check, but, I am pretty certain we do not support a path from 5.5 RC1 to 5.5 GA. I haved tried it in my labs and it did not go well. I recommend that you fully uninstall the VIC, EMS UI, EMS, and nworksMP.

Also, I had noticed in my lab that I had to delete some misc. sites under IIS 7.0 in order to get the EMS UI to reinstall. Check it out and let us know how things go.

Thanks!

-Brian

[krowczynski]

Hi,

just tried to upgrading, but get an error, while trying to import the new nworks MP.
So I tried to, delete the RC1 MP, but get an information, that the MP, had some dependencies on other MP.

So have I delete the other, before I can delete the nworks MP??

[Alec King]

Hello Arkadiusz,

You are correct, you do have to delete the RC1 MP before importing the released 5.5 MP (this is because we made some changes to Monitor types, so the RC will not upgrade to GA MP in-place)

And if you have 'dependencies' from nworks RC MP to other MP's - this is probably because you created some overrides, or some new Groups for nworks objects, something like that. These overrides or groups are stored in some other MP, and they are dependent on the nworks MP. This is why SCOM will not allow you to delete the RC MP.

Note - if you do create overrides, it is best practice to create your own 'overrides MP' for each MP that you customise. Don't use the Default MP for storing all overrides for all MPs.

Which other MPs are listed as having dependencies, when you try to delete nworks RC1 MP?

Thanks,
Alec