1 0 Ref. F1 - VMNCP : OPERATORS' DOCUMENTATION (Ref. F1) + _____ _________ _____________ ___ __ 0 Version 2 + _______ _ 0 8 March 1982 + _ _____ ____ 0 P M Girard + _ _ ______ 0 PART A : GENERAL DESCRIPTION + ____ _ _______ ___________ 0 1. Introduction + ____________ 0 VMNCP is a version of the Network Control Program (NCP) that runs under CMS in its own virtual machine. It can run at the same time as the MVT version, DKNCP, and the two have no internal contacts with each other: effectively they are two separate host machines on the network, each having its own network address. In due course, VMNCP should supersede DKNCP completely, but the latter remains necessary so long as access to HASP is needed, and so long as we have to support some old protocols for the benefit of certain external users. 0 VMNCP will provide network access to VNET (RSCS) for remote workstations, and to CP/CMS for terminals. Terminals will also be able to access MAST by 'DIALing' the FEM, once contact with CP has been established, but this facility is not yet available. VMNCP also supports other applications such as the Bootstrap facility for GEC 2050s, and File Transfer (FTP). 0 The function of VMNCP can be summarised as follows: it is to act as an interface between the IBM central computer complex and non-IBM external networks. In the first instance, it will be connected only to the SERC/NERC Network (SERCnet). Access to the British Telecom PSS Network is via SERCnet to a GEC 4065 'gateway' machine located in the Telecomms area in R26. - 2. The Connection to the Network + ___ __________ __ ___ _______ 0 VMNCP has its own line to the GEC 4080 Exchange. This is a 48K link with virtual address 024. The line level protocol used on this line is the same as the protocol used by HASP and DKNCP on their own lines, namely IBM Binary Synchronous. The data inside the Binary Synchronous blocks consists of X25 packets, X25 being the internationally agreed protocol for making calls across wide-area packet-switching networks. The British Telecom PSS network also uses X25. - 0 VMNCP : OPERATORS' DOCUMENTATION PAGE 1 1 0 Ref. F1 0 The line can of course carry many network calls of various types simultaneously. It should be noted that interactive terminal calls are quite independent of workstation calls, and do not go through VNET even in the case of terminals connected via active networked workstations. The main console of a workstation is usually able either to participate as console in a call to VNET, or to make a separate interactive terminal call in its own right. 0 VMNCP is capable of driving multiple physical lines. However, only the single 48K line is likely to be used in the near future. - 3. Other I/O Devices used by VMNCP + _____ _ _ _______ ____ __ _____ 0 VMNCP writes statistics and logging information to a CMS- compatible minidisk at virtual address 201. Separate utility programs (described later) are available for printing off this information at any time. In the event of an un-recoverable I/O error on the disc, the operator will be informed by a message repeated at intervals on the console. However, the program can continue to function without the disc, and meanwhile an Incident Report should be completed so that someone can investigate as soon as possible. Such an event should be exceedingly rare. 0 The table of network addresses is held in a CMS file called VMNCP TITLES on the VMNCP 'A' disk. This is accessed once only during VMNCP initialisation, but the program cannot run without having read it successfully. See Section 6 in Part B of this paper for further information. 0 VMNCP uses its virtual machine console for communications with operators. This communication is kept to an absolute minimum, and in practice VMNCP can share a physical console with VNET or OPERATOR. If a line restart or disconnection occurs, or if the level of ordinary line errors seems to be excessive over a period of time, VMNCP will send an appropriate message to the console. The program should be allowed to continue running, but investigation is needed. Line restarts or disconnections ought not to occur unless there is a known problem with the 4080 exchange, so it is desirable to produce an Incident Report for these. Excessive line errors may be investigated further by printing off the NCP log (which records all line errors), and the Telecomms duty officer should be informed if there are grounds for suspecting hardware problems. 0 The only commands to VMNCP likely to be useful to operations staff are the TERMINATE and FLUSH commands. - VMNCP : OPERATORS' DOCUMENTATION PAGE 2 1 0 Ref. F1 0 The TERMINATE command is for terminating the program, and FLUSH is for writing the current logging buffer to disc. Both are described later. - - - - - - - - - - - - - - - - - - VMNCP : OPERATORS' DOCUMENTATION PAGE 3 1 0 Ref. F1 0 PART B : NORMAL OPERATION + ____ _ ______ _________ 0 1. Loading + _______ 0 VMNCP is loaded automatically by the autologging mechanism. It can run disconnected, but it is preferable for it to run connected, sharing a physical console with another machine. Initially it will share a console with OPERATOR, but the VNET console may be more appropriate in due course. If it is run disconnected, warning messages about excessive line errors, etc, will of course be lost. NCP is loaded by means of an EXEC file with filename NCP. - 2. Back-up Versions + ____ __ ________ 0 The back-up version may be obtained by re-naming the EXEC file NCP in the VMNCP virtual machine to NCPNEW and the EXEC file NCPOLD to NCP. - 3. Termination + ___________ 0 Wherever possible, it is important to terminate the program by inputting the command TERMINATE on the console. This causes clean termination of all calls and interfaces, and also leaves the virtual machine in a usable state. An 'abend code' of 0000 is recorded on the console. 0 If for some reason TERMINATE cannot be used, the machine may be simply logged off, but this is much less satisfactory from the users' point of view, and also some logging and statistics will be lost. 0 The program should NEVER be terminated by means of the HX + _____ command, as this will leave the virtual machine in an un- usable state, confusing to operators as well as users. It will then have to be logged off anyway. - 4. Operational Control + ___________ _______ 0 VMNCP is designed to run essentially without operator involvement, and no control facilities (apart from the TERMINATE command) are provided at the present time, except for diagnostic and development purposes (see the section on diagnostic aids). - 5. The NCP Log and Statistics + ___ ___ ___ ___ __________ 0 There is enough room in the log for several days under normal running conditions. When diagnostics are being run, however, the log will fill up very quickly. - VMNCP : OPERATORS' DOCUMENTATION PAGE 4 1 0 Ref. F1 - When the log is full, VMNCP merely begins to overwrite it from the front, and the intention is simply to let this happen normally. Thus, the log will not usually be printed unless there is reason to want to look at it, and there will be a limited time during which the data remains available. 0 Facilities for printing part or all of the log at any time are provided by a separate procedure (described later), and it is also possible to print an index of the log's contents. 0 Daily and weekly statistics are kept, and retained for two weeks. It is thus possible at any time during the current week to print off the last week's statistics as a whole. Analyses are made by day of the week, by time of day, and by protocol, and totals are produced for calls, packets, segments and bytes. - 6. Abnormal Termination + ________ ___________ 0 The only abnormal termination code that should occur for hardware reasons is the code 2649. This indicates a fault while trying to fetch the network address table from the CMS file VMNCP TITLES during program initialisation. 0 The program cannot run without having read this file successfully. A copy of the file is held for back-up purposes on virtual machine PMG's 'A' disk. - - - - - - - - 0 VMNCP : OPERATORS' DOCUMENTATION PAGE 5 1 0 Ref. F1 0 PART C : SUPPORT PROCEDURES (EXECs) + ____ _ _______ __________ _____ 0 1. NCPLOG + ______ 0 This procedure can be run at any time from any virtual machine. It 'links' to NCP's logging minidisk and prints off the last 1000 lines or so of the NCP log. 0 It may alternatively be supplied with an 8-digit parameter in the form 'aaabbbnn', where aaa, bbb are the starting and ending block numbers required, and nn is the maximum lines of output (in hundreds) that the user wishes to + ________ print. It will be noted that the user cannot ask for more than 9,900 lines of print. The beginning and ending block numbers must be in chronological sequence, which means + _____________ that as often as not aaa will be a larger number than bbb. - 2. NCPINDEX + ________ 0 This procedure can be run at any time from any virtual machine. It prints an index of the NCP's log, giving the date and time of the start of each disc block. 0 If the user wishes to run NCPLOG with a parameter and does not know the current disc contents, the NCPINDEX procedure provides a useful preliminary facility. - 3. NCPSTATS + ________ 0 This procedure can be run at any time from any virtual machine. If used without a parameter, it prints the weekly statistics for the previous week (Monday to Sunday). If the parameter 'C' is supplied, it prints the current week's statistics (Monday to the present time). - - - - - - - VMNCP : OPERATORS' DOCUMENTATION PAGE 6 1 0 Ref. F1 0 PART D : DIAGNOSTIC AIDS + ____ _ __________ ____ 0 1. Use of NCPLOG + ___ __ ______ 0 If VMNCP seems to be in trouble, or if it has put out warning messages implying trouble on the line, the NCP log should be printed off. 0 Prior to doing this, the command FLUSH should be input on + _____ VMNCP's console, as this causes the current buffer of logging information still inside VMNCP (which may contain up to 73 lines of print) to be written to disc. It also puts out a row of asterisks marking the position. FLUSH is not necessary prior to a TERMINATE command, as the latter automatically invokes the flushing mechanism. 0 For instructions on the use of NCPLOG, see the section on Support Procedures. 0 It is desirable to print the log in the event of trouble, whether or not the operator intends to look at it himself. Otherwise there is a risk of the relevant information being no longer available when someone else takes up the investigation. - 2. Network Status Machine + _______ ______ _______ 0 VMNCP cannot as yet make outgoing terminal calls. However, the Network Status Machine can be accessed via DKNCP or via any other SERCnet machine. From DKNCP, it + __ can be accessed by typing ++N !!NET. 0 At the present time, the Network Status Machine does not automatically monitor VMNCP, and the message it puts out about communications being OK or not refers to DKNCP. In due course this should be remedied. - 3. Making Network Calls + ______ _______ _____ 0 The easiest way of finding out if VMNCP is in a healthy state is to make a terminal call to CP/CMS through it. 0 This can be done from any terminal having access to DKNCP + __ or to another machine such as a 2050 or a 4080 MUM. 0 The network address of CP/CMS via VMNCP is RLIBI, so (if MAST and DKNCP are running) one could for instance type ++N !!RLIBI on the MVT main console, and the reply 'VM/370 ONLINE' should be received if the call gets through. The - VMNCP : OPERATORS' DOCUMENTATION PAGE 7 1 0 Ref. F1 0 call may be cleared by typing ++N @Q. If a reply of '?' is obtained, this probably means that a new call is being attempted without first clearing the previous one. 0 The above test, if successful, also implies that the 4080 exchange is functioning, as the call must have passed through it. - 4. PSS Problems + ___ ________ 0 Neither VMNCP nor DKNCP have any direct involvement with PSS, because PSS (unlike the old EPSS network, defunct as from 1/8/81) is connected to SERCnet as a whole (through the GEC 4065 gateway in R26) rather than to the IBM complex. This is therefore not the place for full instructions concerning PSS, but the following comments may be useful. 0 If someone reports problems in accessing the IBM machines via PSS, the first thing to check is that VMNCP and/or DKNCP are healthy (if the caller is using RLIBI or DTE address 2, he is trying to access VMNCP; if he is using ELEC or RLIAI or DTE address 1, he is accessing DKNCP). Ways of checking VMNCP's viability have already been described. If the NCP and the 4080 exchange are apparently in order, the problem may lie in the 4065 'gateway' machine in R26. If this appears to be working also, the trouble is probably in the PSS Network itself. By telephoning 01-928 5911 or FREEPHONE 2473, the PSS recorded information service may be contacted, to see if there is a known fault. Failing that, the PSS Network Management Centre (NMC) may be contacted on 01-928 5810/5819 or FREEPHONE 2125, though engineers may not be available there at all hours. - 5. Abnormal Termination Dumps of VMNCP + ________ ___________ _____ __ _____ 0 If VMNCP terminates abnormally, a dump is produced and the abend code appears on the console in decimal. For instance a code of 0196 corresponds to the more familiar X'0C4'. 0 The PSW and registers quoted at the front of the dump should be ignored, apart from R0. R0 points to the STAE work area, whose contents are as follows:- - - - VMNCP : OPERATORS' DOCUMENTATION PAGE 8 1 0 Ref. F1 0 Word 0 : Zero Word 1 : Abend code Words 2,3 : PSW within STAE exit Words 4,5 : PSW at time of abend Words 6 ff: Registers 0 to 15 at time of abend - 6. Technical Documentation + _________ _____________ 0 Much technical documentation now exists describing the protocols used in SERCnet, and the internal functioning of VMNCP. For this, see the Index (Ref. I1). - - - - - - - - - - - - - - - VMNCP : OPERATORS' DOCUMENTATION PAGE 9