EOS MGM crashed if it does not meet forward-confirmed reverse DNS

On my site, if rDNS(MGM's IP address) == DNS(MGM's FQDN), eos@mgm.service crashes (the status become inactive(dead)) when an EOS client connects to the MGM node. As a result, the client become unresponsive just after printing the copyright notice.

  1. Is rDNS(IP address) == DNS(hostname) actually a mandatory requirement?
    Otherwise, is this my misconfiguration at irrelevant place?
  2. If it is mandatory, could MGM send error messages to clients instead of shutting the MGM service down?

I have no idea which log is relevant or useful for debugging, any suggestions would be greatly appreciated.

Hi Masanori,

There is the requirement that the IP address that the MGM gets when it starts up (got by doing getaddrinfo) matches the public IP address that the client sees. This is needed for some query commands like xrdfs spaceinfo which have a more involved mechanism and impose this requirement.

Now, having the MGM crash just due to a client request, is not normal. Therefore, could you post a stacktrace of the MGM when this happens. You probably already have something in the logs just before the crash. What version of eos are you running?


Hello Elvin,

Thank you for your response. I understood the background now.

My institute had corrected the registration last week and I failed to reproduce the incident using local resolver with artificially bad rDNS settings now. I will search for logs.

The version information is pasted below:

# ---------------------------------------------------------------------------
# EOS  Copyright (C) 2011-2020 CERN/Switzerland
# This program comes with ABSOLUTELY NO WARRANTY; for details type `license'.
# This is free software, and you are welcome to redistribute it
# under certain conditions; type `license' for details.
# ---------------------------------------------------------------------------