Good evening. My name is Skip Hanson and I am the Consultant User Interface Designer for NetWorker. I am
reponsible for all things NMC. I started this role just after 7.3 was released and have been working towards a
more accessable, powerful and flexible user interface.
My question to all of you is: What is it in NMC that drives you crazy on a daily basis ?
It could be something very simple. Come on...I know there is something you hate ? If I hear it enough, I will get
it fixed. Which brings me to my next point. We need your help to make NMC and all of NetWorker more usable.
I talk to many of you at different times of the year but I want more. I guess you could say I am bit obsessed with
making things better.
"The GUI is the system."
been using NetWorker for a few years. Even prior to the 'new' NMC.
In our environment we are currently using NW 7.5.1, having recently upgraded from 7.3.3:
NetWorker Management Console version 3.5.1.Build.269
based on NetWorker version 7.5.1.Build.269
I do like the NMC console, especially the later versions, sort of.
When usung NW 7.3.3, I sneaked an NMC from NW 7.5 on to my desk, and got a whole lot extra benefits.
However, after our 7.5.1 upgrade there is one specific issue that I have (got a SR open to include this):
Under Devices/ Libraries/selected library, when I select 10 volumes, right click select inventory - tick "Force load and verify labels" - GO....
the system loads all the volumes into a single drive, and it takes ages.
Sometimes I will do this to 100 tapes, and I'll be waiting all day. I want to (have the option to) inventory 1000 volumes!
Previously on my 'sneaked' NMC, I could do the same, and it would spread the mounting across all available drives (24 in each library), hence the performance of this operation has gone backwards.
I realise that depending on what drives are clicked in the device pane, they can be selected from the "devices to use" list. Is it possible to make all available devices available for this action.
I also understand that consideration needs to be taken into other actions, like labeling tapes, due to drive restrictions with certain pools.
Also, NMC seems a little bloated (also better than previous, with more functions). Do you have any advice on performance, and resource utilisation, as I want to avoid NMC using resources on the NW server, that is already creaking under load. Much though I tell people to install the NMC on their desktop machine, I know people that will only use it when they TS onto the server!
What drives me crazy, PERFORMANCE. Ever since we went to version 7.4.4 Build 650 it is vary slow. Opening a group window to check its status takes a lot longer than erlier verisions. By time the window opens you forgot what you were looking for. Many times I need to use nsrwatch and nsrjb or tail -f deamon.log to see what is going on, because if you wait for NMC to update the activity is over.
When creating a library using NMC you can not specify a name. NMC names the library its scsi path. This keeps the cumbersum jbconfig command neccessary. Fix this and we can finally not have to use jbconfig.
A nice to have that I would like to see envolves licensing. The license window is a nice list to see what licenses we have, but it would be helpful if it could show what server (or client) is using the license.
I'm jealous because you are not having to run NetWorker on Windows platform. Not fair!
Anyway, as for the performance in 7.5.1, I cannot say that it is better than 7.5 or 7.4.x versions of NMC.
Eariler I mentioned that some admins like to TS onto the NW server. I am considering removing the GST server from the NW server, and installing on a Storage Node that is less utilised, and forcing them not to log into the NW server.
However, can you explain whether you are using the NMC installed on your personal desktop, as that will send lots of info to/ from the GST database. Or are you using the NMC on a remote system, closer to the GST database.
I'm not sure if it the current version of NMC, but occasionally the NMC freezes and does not update - close and restart required. Case open with support, but they do not consider this high importance, so not expecting to be fixed soon. I have a copy of the latest NW 22.214.171.124 cumulative fix, which has a few pertinent fixes for our environment. It will take ~ a month to get through our internal process to put into production.
Good discussion !
NMC perfromance could be determined by multiple factors.
First one is, the data transferred to NMC is arriving from nsrjobsd. If nsrjobsd is too busy serving other requests, than a perfromance impact could be observed. We have changed the behavior in NetWorker 7.6 where NMC now recieves the information both from nsrd and nsrjobsd, we are seeing a substantial imporvement in NMC perfromance.
Other bottleneck would be the network itself, NW servers that are busy moving I/Os and acting as storage nodes are prioritizing the I/O movements over anything else. If the network link going to the server is saturated, NMC perfromance will decrease.
Other issues may include OS behavior, we have seen a strange Solaris 10/VxFS 5 errors that actually froze NMC/nsrwatch or any other stdout output for periods of 5-15 minutes randomly during the day.
As for sizing rules we can recommend only 2 rule of thumbs:
1. If you have more than 50 clients in the env. install NMC on a different machine (could be virtual)
2. Calcualte 256MB of memory for every 100 managed clients.
This makes sense, the NetWorker server that show the slowness each have over 550 client all feeding in on two GigE network connections per server. This is a major bottle neck. I have been working for a year with our network engineering group to improve this, but we are still 4 months away from getting the 10GigE backbone running.
When you say 50 clients, I assume you mean NMC clients not backup clients?
50 backup clients We assume 3 client definitions per client which makes 150 client entries within NMC.
Thanks to Skip, with the new NW 7.6 we have a "find" and "sort" functions that will allow one to easily navigate to a specific client/comment/saveset/log entries all on a single screen...it looks amazing, can't wait for it to GA and hear feedback !!
If I load 20 new tapes in to my loader, I have to set a label for them: "a new tape"
which I then have to go mark as recyclable one tape at a time...
Perhaps there is a better way to do this, but I would like the option to:
1) when labelling tapes in the gui, set the "recyclable" flag at label time
2) Set multiple tapes at once as "recyclable" (can be multi select, or a slot range, like when doing inventory)
Also, the default, out of the box install of nsr 751 server on solaris doesnt set up the GST backup correctly - I always see this error:
--- Unsuccessful Save Sets ---
* nsr-server:NMCASA:/gst_on_nsr-server/lgto_gst 1 retry attempted
nsr-server:NMCASA:/gst_on_nsr-server/lgto_gst: No full backups of this save set were found
in the media database; performing a full backup
* nsr-server:NMCASA:/gst_on_nsr-server/lgto_gst savepsm.sh: Command not found
shouldnt this work out of the box without any tweaking?
thanks for asking !