Multiple autochangers on a single server/storage node

Thanks for the response.

From the results of my google searches I gathered not many run multiple libraries. I've done it before but I had them on separate storagenodes. I'm only doing it as a transition this time, but I need overlap for a while. I have two physical libraries each 50 slots. My AC enablers are for 128 and 64 and verified each of them. Looking at the properties and info tab, the original library has the 64 slot license and the new library has the 128 slot license so they are not competing for license resources.

Just seems that nsrlcpd is not to be able to manage both of them at once when they are both enabled.

Edit:

Found this on the nsrmmgd man page:

Multiple nsrlcpd daemons (one per enabled jukebox) may be started and controlled by nsrmmgd to handle the lower-level control of, and interface to, the various jukeboxes. The nsrlcpd processes that nsrmmgd manages may be distributed across multiple hosts, since nsrlcpd runs on the host that the jukebox is on.

When I enable both libraries I only see one nsrlcpd daemon running, from this there should be two, one for each library.

AFAIK, there is no limitation with respect to the number of jukeboxes as long as you have at least the appropriate quantity of enablers.

You should be able to run more than one library per SN but you will hardly find someone who ever configured that.

  - There should be no problem to control them from the Admin GUI.

  - From the command line you of course must specify the jukebox along with the nsrjb command.

    If not, I guess NW will always take the first one as default.

An AC license must at least cover the number of slots the hardware reports.

Be careful with AC licenses if they are of diffent sizes:

  - If you do not add them in the proper sequence, it might be that NW will try to use a smaller one for a larger jbox ...

    and will of course fail.

  - How to prevent that behavior:

      - Install jbox 1

      - Add the appropriate license

      - Install jbox 2

      - Add the appropriate license

Good luck.

Hello.

Error message makes me think that you have a zone issue. Take a moment to check your SAN config, later... delete both libraries, rescan and configure again.

Hello Chad,

I have not seen any NetWorker limitation on how many autochangers a storage node can support

Please check your NetWorker autochanger licenses.  How many slots does each license support?  Also, how many slots are in each jukebox?

What license is assigned to which jukebox?  You can find this by opening each NetWorker jukebox property, and see which enabler code is assign to it. It should show the enabler code.

From the hardware perspective, how are the jukeboxes and drives connected to the NetWorker storage node?  What type of controller is being used?  How many controllers?  What type of cable?  If the jukebox and drives have a SCSI interface, check what type it is.  Single ended?  Wide?  Differential? Fiber?  Does the cable, connector, and controller all use the same interface type?  Did you check the hardware compatibility matrix to see if all the hardware is compatible and supported?

Describe how each jukebox and drives are connected to the host.  Are they all daisy chained together?  What are the cable lengths?  If there is a connectivity issue, then you should be able to see issues in the operating system logs.  Look for any SCSI or I/O errors related to your jukebox, tape drives, and controllers, as well as any device disconnects.

If only one jukebox is physically connected, does it work ok?

Good luck...

Licensing looks ok then.

nsrmmgd will try to start nsrlcpd for each enabled jukebox. Render the NetWorker daemon.raw, and then take a look at the log file.  You will see nsrlcpd starting...  you should also see it stopping too.  Hopefully, the preceding messages will tell you why nsrlcpd had to stop.

Both Libraries are direct attach SAS. Licenses have been verified. Each library functions fine as long as the other is disabled. With them both enabled nsrlcpd (only one instance running) will stop, state Lib A is no longer managed and will start and Lib B will become ready, then cycle back the other way.

I have a thought and I can't test until a remaining full finishes, I have not restarted networker since adding the second library. I'm wondering if a full restart of all the controlling daemons like nsrd and nsrmmgd with both libraries enabled will allow nsrmmgd to start two instances of nsrlcpd as its supposed to do and manage both jukeboxes. I'll post up results when I get the chance to test.

After Enabling both libraries, stopping networker, then starting networker, both libraries are functional simultaneously. Apparently networker is robust enough to add the library on the fly, but nsrmmgd not robust enough to start a separate child process to manage it. I'm guessing something to do with the way it manages the RPC program space. Anyway a full restart starts two nsrlcpd processes as expected and libraries function as expected. Thanks for the time folks.

Take a look at the server or storage node's daemon.raw, from the time when nsrlcpd starts until where it stops.  It should provide you with insights on what happens leading up to the jukebox being disabled.

Remember to render the log to make it easier to read.

cd C:\Program Files\EMC NetWorker\nsr\logs

nsr_render_log daemon.raw >output.txt 2>&1

or

nsr_render_log -S "mm/dd/yyyy"  daemon.raw >output.txt 2>&1

where -S "mm/dd/yyyy" is the date that you want to starting looking at.