e-VLBI Station Recommendations

These are recommendations for stations participating in e-VLBI observations. They are here to help get the best results out of each science run, and as a central repository for the configuration that we expect the stations to use for e-VLBI. As our e-VLBI capabilities and software keep evolving, we'll keep this page updated with the latest information.

 

  1. Mattermost - Communications between JIVE and the stations during an e-VLBI run is done in a Mattermost text-chat, hosted here. The username for the JIVE operaters is 'jive.operators', and they will create a group chat channel, please join it. If you don't get the invitation, please let us know as soon as possible. Keep an eye on the chat window, so we can warn you in case we detect problems with the signal from your station.

  2. Installed software:
    • For stations using a Mark5 the recommended configuration is:
      • Debian Etch ('uname -r' should show: 2.6.18-6-686)
      • SDK 9.2, which consists of the following packages:
        • streamstor: 9.2.1
        • mark5: 2.2.0
      • Since 2020 jive5ab is hosted as DIY-compilation source code on the JIVE-vlbi github, where all code, documentation, (compilation)instructions, versions, and bug reports can be found. Binary distributions for Mark5 systems have stopped after jive5ab 2.9.1, however compilation on those platforms is still supported.
      • Firmware versions: Using "ss_rev?" should report:
        ApiVersion 12.13 : ApiDateCode Apr 13 2011 (or later).
        See this page for instructions on updating your firmware.
        N.B.: using any other firmware will cause problems with starting e-VLBI
        Make sure that the diskpacks have been erased with SSErase while running this firmware before recording. Otherwise, recording will fail.
    • For stations using a DBBC/FiLa10G:
      We will configure the destination and VDIF data format on your FiLa10G and start the data stream. However, we rely on the you to configure the other settings of the DBBC and FiLa10. This requires multiple connections to the DBBC, to this end we have developed a dbbc proxy, which is available here. Please have this program running and let us know the IP address and port the program is listening on.
     
    • Starting jive5ab - Stations should start the jive5ab software themselves before a session. To give both the telescope operator and the correlator operator access to the jive5a logging output, the preferred way to start jive5ab is using the StartJ5 script, which is available here. This script will log the output of jive5ab to a file in ~/jive5ab.logs/ by default. The log location is editable in the configuration section of the script, at the top of the file.

    • Automated FlexBuff transfers.
      To have your FlexBuff recordings automatically transferred to JIVE, please take the following in account:
      • FlexBuff access - Access through SSH to your FlexBuff is required. To make this possible in a secure way, please allow access to the 'oper' account by SSH from JIVE, by installing the JIVE SSH public key in ~oper/.ssh/authorized_keys
        SSH connections from JIVE will come either from the JIVE IP-range (192.42.120.0/24) or from the Astron firewall NAT address (192.87.1.181 AND 192.87.1.182). Using the 'from' clause in the .ssh/authorized_keys, it is possible to limit access with this key to only these originating IP addresses for added security.
      • .JIVE.allow - You need to create a file called: ".JIVE.allow" in the /home/oper directory. This file gives you the opportunity to make a list of timeslots in which JIVE is allowed to initiate e-transfers of recorded data. Here's an example of a line that is used in many Flexbufs:
        2014-07-01 00:00:00 2042-07-30 00:00:00 0
        The date/time range is self explanatory. The last number indicates the maximum allowed data rate in Mbps, where zero stands for unlimited (1000 = max.1 Gbps). Multiple lines/timeslots are allowed. In case timeslots overlap, the lowest data rate will be used.
      • .JIVE.transffered - In return, after a transfer has finished, JIVE's automated transfer software creates a file on your FlexBuff called ".JIVE.transferred" to which the transferred experiment names and the timestamp when they were transferred, are appended. Experiments listed in this file can be removed from your Flexbuf. JIVE will not remove any data from your FlexBuff, we leave that to you.
      • Recording names - Apart from access to the two files mentioned above, we also need SSH access to scan for recordings. We will scan in the /mnt/disk* directories for recording which start with "<experiment>_<station>_", all in lower case.
      • Jive5ab - Just like in the case of eVLBI, you should have an instance of jive5ab running on your FlexBuff for JIVE to use to transfer the data. This can be the same instance as the one used for recording or a separate one.
      • Firewall - Your firewall should allow TCP traffic to/from our control computer on port 2621 from IP: 192.42.120.116 (jive5ab default listens to port 2620, but stations that do automated transfers usually use a separate instance of jive5ab listening to port 2621).
        Your firewall should allow UDP traffic on port 2630 to 2640 to and from IP range 192.42.120.34 - 46 (these are the Flexbufs at JIVE).
      • Optimization - Notes on how to configure your FlexBuff for optimal recording performance can be found here.

    Last updated: January 2019 (revision 1.4).