NMC - What drives you crazy ?

Hi Larry

Not sure what your issue is with NMC on AIX 5.3 because we have had it working for a while now, we are running NetWorker 7.4.4. and the following NMC version.

/nsr/lgtonmc/bin/gstd:
        23  1.4  src/bos/usr/ccs/lib/libpthreads/init.c, libpth, bos510 8/19/99 12:20:14
        61      1.14  src/bos/usr/ccs/lib/libc/__threads_init.c, libcthrd, bos510 7/11/00 12:04:14
         NMC Release:  NetWorker Management Console 3.4.4.Build.634
         Build arch.:  aix51n
         Build number: 634
         Release:      7.4.4.Build.634
         Product:      NetWorker
         Build date:   Thu Dec 11 17:33:29 PST 2008
         Build info:   DBG=0,OPT=-O2 -fno-strict-aliasing
        57  1.4.1.4  src/bos/usr/ccs/lib/libc/fstab.c, libcenv, bos510 6/7/00 16:35:25

I totally agree with you on your comments regarding how groups are handled under the configuration tab in NMC - why on earth when you click on groups do the groups get displayed in both panes that is annoying...

1 more thing for Eric and the other EMC'ers I decided to give NMC 7.5 a go on my desktop (Windows XP) and on the performance front I see a step backwards - my GUI is now consistently hanging with the progress bar saying 16% for sometimes minutes at a time - a step backwards from the 46% that I was complaining about yesterday ....

Cheers

Pete

Another...

in Monitoring, when sorting by "Next Run", 'disabled' is highest priority, or counts as lowest time. I want this to show what group will start next at the top. If I click "Next Run" again, the last to run group will be top.

Phil

SkipHanson wrote: We need your help to make NMC and all of NetWorker more usable. Cheers, Skip NetWorker Usability "The GUI is the system."

Skip,

just thought of one things that would make NMC more usable - sorting, mutliple column.

When in the Libraries pane, there are 2800 slots, many slots with volumes. I want to sort by Pool, then by %used.

Hope I don't get banned - ECC (Control Center) can do this type of sorting.

Also,

In Monitoring/ Log, could there be a way to set the number of lines to store on the individual users log? This would also be useful in the Operations tab, as we could see what occurred prior to the tape library malfunction/ stuck tape.

Also,

In Monitoring/ Sessions, I have recently been carrying out lots of manual clones. If I leave clones running overnight and want to check the clones first thing the next day, I expect the previous session type to be the default, when I go to the tab. This is not saved. I assume many options/ settings are not worth saving, but this would allow me to login and "be where I was". You could also add a click button to reset GUI settings to default.

Phil

PeteOrton wrote: 1 more thing for Eric and the other EMC'ers I decided to give NMC 7.5 a go on my desktop (Windows XP) and on the performance front I see a step backwards - my GUI is now consistently hanging with the progress bar saying 16% for sometimes minutes at a time - a step backwards from the 46% that I was complaining about yesterday .... Cheers Pete

Pete,

sorry but I'm going to laugh at this situation.

I had similar situation. I opened a case with Support. They told me to ~'go and jump'/ 'that is not supported'. I ignored them, as we were using NW 7.3.3 at the time, and I wanted some 'nice things' like select and label 20 volumes, using point click.

Out of curiosity, did you install the NMC by installing the '7.5 server bundle', and just not having the server backing anything up. Thats the only way I could get the NMC installed on my laptop. This provided lots of communications betweeen the real server and my laptop server, and even more stuff in the logs when I was on-call/ having the laptop off at night/ weekend.

Since the upgrade of the server to 7.5.1, I've uninstalled the software on my laptop, and installed the NMC 'correctly', and the logs do not contain my laptop anymore

Phil

Another,

filter (again in ECC). I could filter the 'disabled' group, as in the previous example. However, We require an icon, next to the operator icon on the right, to show that the filter is in use.

time for a breather

Phil

I could not get something to run, I think it was the GUI. In their defense I was one of the first to try putting it on AIX, in fact it wasn't even listed as one of the operating systems at the time. They did have it packaged and everything for AIX I just could not get it to run. Might have been a Java runtime issue or something along those lines because my DBA was also messing with Java updates at the time. Luckily I had/have a Windows server that gets very, very little traffic so it is not a big deal to put it on there.

I guess the thing was that I just wanted to keep everything in one place.

About the naming of a jukebox. Just disable the jukebox. Open properties and rename the jukebox to what you like, we did on our server running NSR 7.4.4.

Now we have a library named rd= :DR  for the disaster site. Good luck, //Håkan Engman

Hi Mikey,

Must ask you again to just clearify this on memory usage on the NMC server.

Please correct me if i'm wrong.

You recommend 256 MB/RAM per 100 client definitions on the backup server.

Our largest server have 935 uniqe clients, and 1502 definitions.

That would make a recommendation of memory on the NMC to -->> 1500/100 * 256 = 3 840 MB

The NMC server is centralized, so can monitor another 6 servers, don't know the exact number of clients,

but my question is.

When is the memory used/allocated ? Will monitor memory usage and paging, but maybe you know

the facts.

-Does "Capture Events" and "Gather Reporting Data" have any impact on the memory usage ?

A gut fealing is that a bunch of Networkers out there run a NMC server with inadequate memory installed.

Hope we see you in Sweden again sometime soon Mikey,

Regards/Håkan

Well, I guess that we have to live the limitations upuntil re-arrangement of this config.

The thing i wait for the most is the ability to define the controlpath myself, or i.e. get support

for virtual controller targets. In our case HP-UX provide a virtual jb control path pointing to

two physical connections through our dual SAN infrastructure.Loosing the controler path will

make you go out of service after a while, even though it's quite doable to change it manually.

rd=  is the way Networker knows the target is a remote target.

, well informative I think. Where to dig when you get problems...

Don't know what you are scripting, but in most cases you don't need to

specify the jukebox. nsrjb -u would select the proper jb, and

mminfo-pool usage might do the "other" selection.

Cheers,

Håkan

Phil_1 wrote: Hakan_Engman wrote: The thing i wait for the most is the ability to define the controlpath myself, or i.e. get support for virtual controller targets. In our case HP-UX provide a virtual jb control path pointing to two physical connections through our dual SAN infrastructure.Loosing the controler path will make you go out of service after a while, even though it's quite doable to change it manually. Cheers, Håkan I understand, I think. When my first path went down, I just changed the path (scsidev@11.0.0 to scsidev@13.0.0 on Windows ), (kept the name) and it worked. It was on the remote jukebox as well. I would expect this to work via a virtual control path, however, I have not ventured into the virtual path or device.I've been warned off this by support, as well as warned of DPF (Data Path Failover). 'Not tested' is not what I want to hear from the company who boast of the ~'biggest testing environment on earth' - at least bigger than their compeditors. Anyone else done this via Windows/ IBM CPF (Control Path Failover), or DPF.

I do think that robot controller failover would be ok, as thera are no on-going sessions like on tape. Would never try path failover on physical tape, VTL drives should be immune to SCSI resets so it might be ok. Talking about IBM driver for tape and HP-UX driver, they are not completley loadbalancing as wanted.

The choosen path is selected upon start (boot ?) and the drive would continue to use that path until failiure. Resulting in usage of only one HBA in worst case. If the selection where session based, the OS would "loadbalance" over multiple HBA:s if available. But one have several drives, but only one contrlpath..

I guess that this is a limitation in the SJI driver, if it's still around in Networker.

We use "nsrjb -j -u", to unload all drives prior to ejecting tapes. We could get around this by using a variable LL= DR=, but then different people doing different scripts will use different variables presents additional problems. Also, when using mminfo with -r location, . Phil

Environment variables for the libraries seems ok.

Howabout:

mminfo -q "

mminfo -avq "" -r "location" -xc;   and use delimiter(;) when fetching the variables.

Hakan_Engman wrote: The thing i wait for the most is the ability to define the controlpath myself, or i.e. get support for virtual controller targets. In our case HP-UX provide a virtual jb control path pointing to two physical connections through our dual SAN infrastructure.Loosing the controler path will make you go out of service after a while, even though it's quite doable to change it manually. Cheers, Håkan

I understand, I think. When my first path went down, I just changed the path (scsidev@11.0.0 to scsidev@13.0.0 on Windows ), (kept the name) and it worked. It was on the remote jukebox as well. I would expect this to work via a virtual control path, however, I have not ventured into the virtual path or device.I've been warned off this by support, as well as warned of DPF (Data Path Failover). 'Not tested' is not what I want to hear from the company who boast of the ~'biggest testing environment on earth' - at least bigger than their compeditors.

Anyone else done this via Windows/ IBM CPF (Control Path Failover), or DPF.

Hakan_Engman wrote: Don't know what you are scripting, but in most cases you don't need to specify the jukebox. nsrjb -u would select the proper jb, and mminfo-pool usage might do the "other" selection. Cheers, Håkan

We use "nsrjb -j -u", to unload all drives prior to ejecting tapes. We could get around this by using a variable LL= DR=, but then different people doing different scripts will use different variables presents additional problems.

Also, when using mminfo with -r location, .

Phil

Phil,

  Hello. Apparently we support multi-column sorting today. Just keep in mind the order of the clicks.

If I click on the table header for age, then the table header for name you'll get.

Ben       7
Ben       9
Glen      6
Jack      5
Jack      8
Jack      9
Tom      6

Check it out and let me know if this helps ? We need to be more obvious about how this is done. I am

talking to eng about this. This in one of those "simple" things that could make a big difference.

Cheers,

Skip

Something I just noticed. I don't pay attention to it because it is rather usless. The Log tab rolls over too quickly. The Log tab is too small to be useful for us.

Good morning. This is something we have heard a lot lately and I can tell you we will be addressing this VERY soon.

We need to make sure we don't affect the performance of the NW server with any change we make. Thank you

for sending this, it is a perfect example of how things can get changed when we hear it enough from our users. This

is exaclty how this process is supposed to work.

Cheers,

Skip