NMC - What drives you crazy ?

On the use of NMC from 7.6 with earlier Server versions - yes - it's OK to do that.  We discussed this a few months ago that some of the advantages in the new UI were things customers would want to use without having to do a complete 7.6 upgrade.  We simply didn't have a clean way of promoting that you could do this.  As Hrvoje points out, sometimes when you do this you aren't able to use some of the capabilities.  (I can't think specifically of what you wouldn't be able to do - actually I guess the VMware mapping wouldn't work if you were on an environment below 7.5).

I'll let Skip reply on the RFE of having folders in the config screen for groups.  Great idea - and shouldn't be too difficult to implement.

- Eric

I'd just like to add that this one

"Now the bug bear we have is that under the Devices tag under the Devices window, if one our less well trained ops tries to be clever and "Scans for Devices" then lots of irrelevant devices are created. Apart from hacking the res database is there anyway of getting rid of these useless devices?"

Also drives me a little crazy.

Scott

After the "Scans for Devices" is run you will get the drives listed in orange in MMC correct? if so, select the device, configure as a stand alone drive and then you will be able to delete it from the list  by right clicking on the stand alone drive and selecting delete. Hope it helps

Well, what make us crazy about NMC is as follow :

We are currently in deployment process of Networker 7.4.4.  As some as explained, we have very low response time when creating new clients (5 minutes) or updating one. The administration team is really going crazy about that...

We are in 7.4.4 and are not at the point of changing the version yet, so we're kind of stuck.  I have actually two networker servers, one in techno and one in Prod.  The techno server is so fast, it's unbelievable, but the Prod one.... it's so slow...  The difference between the two is the the load, there is much more clients on the Prod than the techno, maybe 1000 compared to 50.  Can that be a reason ?

What can we do to fasten the Production Env ?  Does the fact of installing a newer version of the NMC could help us ?

Thanks,

Christian

Hello, CarrierC.

The best way to start chasing performance problems is to make sure you are amed with knowledge.  Do you have performance information for the box?  Is the bottleneck CPU, or memory / paging, or disk IO, or ???  Until you have graphs of CPU usage, memory and so on plotted through the day, you're just guessing in the dark.

Better hardware is always something to consider, informed from knowledge that you need more CPU or memory or whatever.

You'd benefit from running the NetWorker Management Console on its own dedicated hardware if you environment has more than ~100 clients on the NetWorker server.

As for the software, I'd suggest adjusting how often NMC polls the NetWorker server(s).  Lest I mis-remember where that is set, I'll ask you to look in the admin guide.

Remember that you can also install a NetWorker 7.6 client, and NMC, on some new machine in your environment, and use the 7.6 NMC to control your 7.4 NetWorker server.  That will gain you more than two years of ease of use improvements that Skip and the NMC team have made since 7.4.

Hope this helps.

NMC shouldn't poll the NW server except to make sure it is still alive. the NW server should send data to NMC.

wow, nothing about the reports.

OK, here's some:

When you look at drive utilisation, it shows you all the drive paths. I want to see a physical drive's utilisation so I want a drill down report. See each physical drive and then click on that and then see the various drive paths to that Physical Drive (Go look at my requirements for 3.2 of NMC it was in there).

Another report I want to see is the throughput of a drive - that's physical drive - which can then be drilled down on to show the individual drive paths. It would be good to be able to select speed and/or sessions on the same graph so you can relate the two. Would so help with performance management.

Also, why is it that the user report only refers to changes made in NMC. I want to see what is in the rap.log. NetWorker security and NMC security have to be tightly bound.

This is sort of more related to NetWorker, but also is sort of NMC. I want to be able to set permissions individually on resources, or groups of resources. That is, a user can operate a savegroup, but nothing else.

When looking a jukebox in device Management you should see the physical drives, not all the drive paths. If you want to install to a particular drive, you should then explode the physical drive (or all of them). When in a DDS situation you can have way to many drives. Most the time you are just monitoring and you could care which drive path it is in, just that the drive is being used.

I'll tell what annoys the hell out of me. Being told I have 60% dormant files.... but not being told which ones! ARGGHHHHH! That's like being told you are going to get stabbed tonight, but not by whom, at what time, or where! Bloody annoying!

Some of the reports are completely superfluous - get rid of them!

Have an integrated GUI with AlphaSore, I mean AlphaStor.

Have  a single GUI for Avamar. OK, you want a single stand alone product, but why do we have to learn two UI's? My rant on Avamar and NetWorker "integration" will be another topic.

Why can't I launch my VTL GUI from NMC? Oooh, it's a VTL, right click, launch GUI!

NMC was originally conceived as a single management console for many products... use it like that! To start with, just make it a launch point. If you can't do it, let us add applications easily. That means multiple applications from the same host (That was broken when Retrospect was added)

integrate with your product that find all the stuff out there (forget its name) and enable you to automatically find out if hosts are being backed up or not (Much like you do with VC).

Oh, set up a backup policy on a VC so that if a new VM is added, it gets backed up automatically!

It drives me crazy that EMC says that there is something called client distribution feature in NetWorker (nsrpush / Software Administration service).

But it is NOT an distribution feature in reality. It's merely an upgrade feature.

It would be a great improvment in NMC / NetWorker if there was good way of installing clients and modules on any kind of backup client that previousle does not have NetWorker installed without the need of SMS/ altiris or such expansive and complicated software.

Johannes

I too will have to cite performance as a major issue. We have NMC on a dedicated server but with upwards of 2500 clients defined it's painful to use for anything other than rudimentary monitoring. It's not so much speed of operation, it's time to update. Right click to restart a group, you can wait up to 20 seconds before it goes green. If you need to restart 10 groups, it gets boring quickly.

I tend to do pretty much everything from the command line, on the Networker server.

nwadmin still does a better job at giving a timely at-a-glance view of what's happening.

I think most of the issue goes back to jobsd, but that's a whole other area that I'm not happy with.

We're using 7.5.2 across the board on an AIX / Linux server backend.

Jason.

I am seeing the same performance issues with NMC, I am running NW 7.4.5 and  NMC 3.4.5, we roughly have 1100 clients. I come in and the morning and bring up the NMC GUI and wait for about an hour for sessions to appear, I have stopped Networker renamed the jobsdb and happens a few days later. I use "nsrwatch" to monito. I wish they would bring back the NWADMIN gui.

I am glad to see it hasnt been fixed in 7.5

You may want to see if your jobsdb size is out of hand. It is recommended that it purge on time (3 days) and not on size. If you reduce the time you retain your jobsdb and increase the size it is allowed to grow so it purges on time it will work better for you.

This should really be under it's own thread, jobsd - What drives you crazy?

I have retention set to 3 days and size set to 200MB on one datazone.

jobsd runs at 100% on one CPU 24x7 and at present the jobsdb is 800MB in size

and contains 200k objects.

jobsd has been the single cause of every major problem I've seen since it was introduced,

regularly hanging up, giving up and simply forgetting or failing to schedule savesets until

manually restarted.

Of course with three days retention I lose all detail of backup completions older than that

making group restarts impossible without manual intervention etc..

Jason.

All,

Hello again. It's been a while since I posted any replies. I just got back from EMC World and got a boat load of feedback from customers and TC's. We obviously have issues with NMC performance, both in regards to jobsd and not.

  Based on this feedback we are seriously looking at this aspect of NMC and will be working hard to improve that area of the product in the near future.

  Keep it coming. Thank you all once again.

Skip

I must admit, I've always thought that NMC should receive rather than poll.

Would this not solve some of the scalability issues?