Conference 44.370::system_management

    Graeme, now that the LPS-40 is installed, there should be
    a couple of "spare" line printer about - why not ask Jerry
    or Steve to get one installed on MAIDEN, as a temporary measure.
    
    Re the network, it seems to me that the link to SEPTIC is not
    as fast as it used to be, but I don't have any hard facts on this.
    
    Peter.
    

    The installation of the LPS40 has *not* left any "spare" printers.
    The LPS40 is a different breed of printer (laser/sheet feed/A4 paper) as
    opposed to the impact printers (monospace/continuous feed/132 column
    paper).
    
    Nevertheless, I'll pass this on to Jerry to see what print services
    can be arranged for all you nice PDP people.
    
    steve

    ADGV01's network parameters are the first line of attack. It was
    regenned  and booted last night - any change in respons times Graeme?
    
    The second possibilty is that MAIDEN needs it's DECNET/SYSTEM retuning
    to reflect the new router capabilities.
    
    If this fails to improve the situation, I'll compare the performance
    of printing U39 to SEPTIC as against MAIDEN to SEPTIC which would
    identify if it's a DMR line problem.
    
    Stay tuned for the next installment.

	re Note 74.2    by Steve Draper
    	    -< Printers...wot printers??? >-

    >The installation of the LPS40 has *not* left any "spare" printers.
    >The LPS40 is a different breed of printer (laser/sheet feed/A4 paper) as
    >opposed to the impact printers (monospace/continuous feed/132 column
    >paper).
    
	Funny that, as there has been a band printer (LP25/LP26?)
    	"lying" around in the machine that is not connected to anything.
    	I think it has/had a label on it saying "VULCAN".
    
    I thought (probably wrongly, in the light of your reply) that
    we were going to move from impact to lasers for the vast majority
    of our printing....... and that this would "free up" the impact
    printers.
    
    Peter.
    
        

    Jerry,
    	There has been no improvement with the NETWORK link between
    MAIDEN:: and SEPTIC::. Infact, MAIDEN:: cannot see SEPTIC:: anymore
    and vice versa.
    
    This would point to the Network tables on MAIDEN:: and SEPTIC::
    which seem to need to be updated.
    
    I do not know how to use $NCP, but I will read up on it to try and
    be of some use. Though, would you have any suggestions to get the
    two PDPs connected on the network ?
    
    
    Best Regards, Graeme
    

    Slow DECNET transfers from MAIDEN (and RDGE03) appear to be related
    to hardware errors on the DMR's on ADGV01. Field service are checking
    the error logs.
    
    I've looked at the network Dbases for both nodes, and they both
    have correct entries for all 4 PDP's at Acre Road. It's not this
    that's causing slow response.
    

    The printer labelled VULCAN can't be driven from terminal ports
    such as those on the DHU11's on MAIDEN and U39. It used to run on
    the parallel port of the DMF32 on the old Vulcan Machine.
    I believe there are no suitable ports to drive this printer on any
    of the PDP's or VAXes.
    
    The other limiting factor is that it requires a power supply already
    allocated for the 8530 VAX about to be clustered with U9/U28.

    Jerry, the PDP's can drive parrallel printers using either the
    Centronics or DataProducts protocol if they have a parrallel
    printer port (like wot SEPTIC and RDGE03 have).
    
    Nice printer, shame about the power......
    
    Any one got 50p for the meter???
    
    Peter
    

    Jerry,
    	Help. Its is getting quite bad. I keep running up and down stairs
    to print the files on the console terminal (which is slow).
    
    	Could you express an urgency with the fixes to ADGV01:: ?
    
    
    Graeme
    

    To print stuff from MAIDEN on the main printers on the ethernet
    do as follows:
    
    RUN $NFT
    NFT> NODE ADGV01::
    Node:     ADV01
    User:     MAIDEN
    Password: Printer
    Account:  <CR>
    NFT>     ADGV01::*.*=<your-maiden-file-spec>
    
    This will put your file onto ADGV01's disc. From there it gets copied
    to either the LPS40 (when it gets fixed) or one of the LP25 impact
    printers. The printing job runs every 5 minutes, so there may be
    a 'slight pause' before the job emerges.
    
    This won't solve the response problem of PDP-to-ADGV01 communication
    but it does provide an alternative printing route for PDP's.
    
    I've had no luck yet getting further help with what is wrong with
    the DECNET response to the PDP's. IS Networks don't have the required
    expertise, and CSC Basingstoke will not support the version of DECNET
    on the PDP's.  
    
    Anyone know where I can get a 2nd Unibus for an 11/750?
    	
    

	re Note 74.10 by FOOT::THOMPSON
    
    >I've had no luck yet getting further help with what is wrong with
    >the DECNET response to the PDP's. IS Networks don't have the required
    >expertise, and CSC Basingstoke will not support the version of DECNET
    >on the PDP's.  
    
    This could be a little embarrassing, as this is the set-up what
    we use for all of our internal PDP's, you know, the ones that are
    supporting about $3,000,000,000 worth of business in Erope.    
    
    Sigh. 
    
    Peter.
    

    Peter, if I could get my hands on some of the $30scrillion then
    I could come up with a better solution. But I can't, so I have to
    work within the constraints of unsupported software versions and
    the hardware that I can beg/borrow/steal from DEC Park.
    
    I believe slow response is due to the UNIBUS on ADGV01 being overloaded
    with 4xDMR11's, a DEUNA and a UDA50. I'm getting a 2nd
    Unibus installed so the load can be more evenly distributed. 

    What ideas have you got to make it go-faster?

    My suggestion? You won't like, but put some of the PDP's
    onto some other VAXen.
    
    To prevent loading the systems too much, why not share the load
    by putting one PDP on each cluster, or each VAX?
    
    The PDP network situation is not acceptable, and it is impacting
    workloads.
    
    I sympathise with the obsolete hardware/software you use, but I
    use it too..... all the time!!
    
    It wouldn't cost anything (except FS time) to revert back to
    what we had before, and it wouldn't be worse than it is now.
    
    (Thinks : maybe this discussion would be better off off-line.)
    
    Peter.
    

    Current DECNET problems with the links to MAIDEN appear once again
    to be related to a faulty DMR11/interface within ADGV01. FS are
    checking the hardware.
    
    

    A *new* LP27 parallel printer will be installed on SEPTIC shortly
    to replace the aged LP11. I believe an LP27 is a bit faster than 
    the LP11 (up to 1200 lines per minute) and according to Field service 
    LP27's are very reliable beasties.
    
    The current LP11 will be scrapped, and once the LP27 has been running
    for 2 or 3 days the LP25 printer on U3 will be decommisioned and
    suffer a similar fate. 
    
    This means that U3,U39 and MAIDEN have no printers attached to them
    and print jobs must be sent by DECNET either to SEPTIC or a VAX
    for laser printing.

    As readers of this note will be aware, DECNET response from MAIDEN
    in particular is not good. Other descriptions of the response have
    been used, but I'll stick to "not good" for this note. As part of
    the continuing saga of trying to improve the response to MAIDEN
    I'd like to shut down both MAIDEN and U3 from 12:15 to 14:15 tomorrow
    (14th January) for more tests on DECNET.
    
    If this is inconvenient please let me know asap - 6746

    ADGV01 has had a few hardware changes made today. It's had 3Mbytes
    of memory added and the 4 DMR11's have been shuffled around in the
    backplane so that they're all in the same cabinet, separate from
    the UDA50 & DEUNA. Also, the number of buffers on the 'U3' DMR link
    has been reduced.
    
    Before all this was done, the transfer rate between U28 and Maiden
    was typically 12 blocks in 2 minutes. It's now sending 2000 block
    files between the 2 nodes in under 10 seconds (elapsed). 
    
    The same transfer rate appears to apply to the other 3 PDP's as
    well.

    The LP27 replacement printer for SEPTIC's LP11 is being installed
    on Sunday 24th January.
    
    U3's printer will be deinstalled when the DECnet router problems
    are solved.

    and it is working fine, too!!!
    
    Peter		(with something nice to say for a change!)