File: /usr2/fs/misc/m5/readme.txt WEH 20020412 Support for M5 in FS 9.5.7 This document is divided into five sections: 1. Software, decribes the software commands used with the Mark 5 recorder, 2. Caveats, describes warning and issues you should be aware of 3. Operations, describes how the system is intended to used operationally 4. Hardware, describes what the supported hardware configurations are 5. Installation, describes how to install the new syste, SOFTWARE There are a few new commads: "disc_start", "disc_end", "disc_pos", "disc_check", "mk5", "mk5close", "mk5relink". These are documented in the style of SNAP command manual pages at the end of this file. There is one new standard procedure "ready_disc", documented at the end of the file after the new commands. There is one new control file, "mk5ad.ctl", also documented at the end of this file after the new procedure. The usual "echo=on" mechanism can be used to record the raw communication strings between the FS and Mark 5. String sent to the Mark 5 are shown in square backets "[]"; string read from the Mark 5 are shown in angle brackets "<>". As usual commands that encounter no error and have no data to log do not show a response unless extended display ("xdisp=on") or extended logging ("xlog=on") are turned on. CAVEATS The Mark 5 support has not be tested extensively, but it seems to be reasonably robust. Please report any problems to weh@vega.gsfc.nasa.gov. If you specify a Mark 5 device in "mk5ad.ctl" file and a initial connection cannot be made, the FS will not terminate. This means that you can leave the mark 5 device defined in the control file even if it is truned off or otherwe unreachable. After an initial failure to connect, the FS will try to establish a connect for each subsequent Mark 5 command (except "mk5close") until a connection is made. It is important to specify an IP address for the Mark 5 in the "mk5ad.ctl" as opposed to a host name. If a host name is used and there is some problem with the domain name server it wil be impossible to connect to the Mark 5. Another possible problem is that for some network problems, it is impossible to make the connection to the domain name server "time-out" and abort in a short amount of time. In this case the FS will hang for a substantial amount of time. The later risk is unlikely to occur and in fact many common problems are handled well by the time-out mechanism. However, both of these risks are, as far as we can tell, completely eliminated by using an IP address. It may also be possible to eliminate the risk by placing the mark 5 hostname and IP address in the /etc/hosts file, but this hasn't been tested. If a host name is and the connection attempt hangs and doesn't time-out, the only solution is to make sure the window focus is in the window you started the FS in and press Control-C to cause it to die (sorry about this, but don't use host names if you want to avoid it). After a connection is established, the FS is able to detect time-outs when communicating with the Mark 5 except possibly in some pathological cases that haven't been encountered yet. If the Mark 5 times-out after a connect has been established, it will attempt to re-connect to it. This should work well, but if you get a hang, use the method described above and contact Ed (weh@vega.gsfc.nasa.gov) with as much information about what was happening as you can. Some mode limitations are also described in the Operations section. OPERATIONS The Mark V recorder is currently supported in a "piggyback" mode only for test observations with the Intensive UT1 measurements between Kokee and Wettzell. To use the piggyback mode, utilize option (13) in DRUDG to tell it that you want this mode. DRUDG will then insert the necessary commands into the *.snp file for the experiment. No other changes are made for back-ends with VLBA formatters. For Mark IV systems, the "trackform" set-up is modified in the *.prc file to write the same data to the head 2 output of the formatter as is going to head 1. The piggyback observations are fairly straightforward. The disc is started ("disc_start=on") and stopped ("disc_stop") along with the recording of "valid" data on the tape. The disc position is recorded ("disc_pos") in the log before each start and after each stop. After each observation a simple check ("disc_check") is made that data was recorded. No error message is issued if the check finds no data, but this provides a record of what was or was not found on the disc for subsequent trouble-shhoting. The only other difference in that the "ready_disc" procedure is invoked by the schedule to: disable the link to the Mark 5 ("mk5close"); prompt the operator to insert the discs (As part of inserting the discs. the operator may need to terminate the Mark 5 recorder's control program and restart it. This can be done via the Mark 5's console or an "xterm" logged into the Mark 5, wherever the control program was started from. Please check with any documentation provided with the Mark 5 on its operations for more details.); after the discs are ready, the operator enters the SNAP command "mk5relink", the connection to the Mark 5 is resestablished; the disc serial numbers are recorded ("disc_serial") in the log; and the byte position ("disc_pos") is recorded in the log. If it is impossible to re-establish the link to the Mark 5, the schedule can be continued without the Mark 5 by entering the SNP command "cont", but in this case all the subsequent Mark5 command will generate errors until it is possible to establish a connection. Operation of the Mark 5 recorder itself isn't covered here in any detail. It should be covered in any documentation, training, technical assistance that is provided with the Mark 5. However, a few points are summarized here with emphasis on points of interaction with the FS. The control program for the Mark 5, "Mark5A", must be started for the system to be operational. This can be done either on the Mark 5 console (if one exists) or by logging on to it from an xterm. The program is started running in background by entering "Mark5A &". If you want debug output, put a zero "0" before the ampersand, as in "Mark5A 0 &". When it says "ready", it is okay for the FS to be started. To terminate the program, enter "EndM5". When it says the "the end", it is free and clear. This can be done while the FS is still running and if the program is subsequently restarted the FS will reconnect on the next Mark 5 command that is executed. It may be necessary to stop the program before removing discs and restarting it after inserting discs. The FS and the Mark5A program will operate more cleanly if you close the FS connection (with "mk5close") before stopping the program and re-establishing a connection (with "mk5relink") after re-starting it. The "ready_disc" procedure uses this method. One issue that is not resolved at the time of the writing of this document is how "erasing" disk will be handled. There are arguments for it being done either at the correlator or at the stations. In the event that it needs to be done at the stations, the command "mk5=reset erase" can be used. This will cause all data recorded on the disks to be unrecoverablely erased. It should not be used as a normal part of a schedule, but only to re-initialize the disks after they are initially received at the station. Extreme care should be exercised when using this command, since it is equivalent of degaussing a tape. The possibility of an inadvertant erasure at a station is one of the arguments for this operation being done at the correlator before discs are shipped. Currently only modes that use "trackform" to set-up the formatter are supported. In other words, simple Mark 3 modes are not supported. Another temporary mode restriction is that only modes that record all 32 tracks can be used. Any mode can be made to do this by assigning data to any unused tracks. HARDWARE Currently only two hardware configurations are supported: (1) VLBA formatter in which case the Mark 5 recorder is plugged into output for recorder 2, and (2) a Mark IV formatter, in which case the head 2 outputs should be unplugged from the recorder and the Mark 5 should be plugged in. INSTALLATION If you are still using a version earlier than 9.5.3, you should upgrade to 9.5.3 first. Please contact Ed (weh@vega.gsfc.nasa.gov) if this is a problem doing this. If you are using 9.5.3, it is fairly painless to install the new version. (1) Place the archive fs-9.5.5.tar.gz on /tmp. (2) Stop the FS (3) As "root", execute: cd / tar -xzf /tmp/fs-9.5.5.tar.gz cd /usr2 ln -sfn fs-9.5.5 fs (4) As "prog", execute: cd /usr2/fs make (5) When the "make" completes, please type "make" again. If no errors appear after this "make", everything was successful. (6) Set-up the "mk5ad.ctl" control file. As oper: cd /usr2/control cp /usr2/fs/st.default/control/mk5ad.ctl . Edit the file "/usr2/control/mk5ad.ctl" and enter the address, port, and time-out for the Mark 5 recorder. Using the IP address instead of the host name is strongly recommended. If you only have the host name, it is easy to look-up its address with "nslookup". The initial port number is 2620. It is unknown what the correct time-out should be. Initially, 100 is recommmended if the Mark 5 is on the local network; 500 for a remote device. If time-outs occur these values can be increased. (7) Append "ready_disc" to your station.prc file. As "oper": cd /usr2/proc cat /usr2/fs/st.default/proc/m5.prc >> station.prc (8) As "oper", try the FS. It should be available for use immediately. There should be no need to reload station software or reboot. ---------------------------------------------------------------------------- disc_check - check Mark 5 recorded data Syntax: DISC_CHECK Response: DISC_CHECK/,