Hi,

Since the upgrade to 2.0.0 I notice that the known_hosts property is
no longer getting populated by the hostname of the server to which
I've connected. This was quite useful to us for diagnostics since we
use a proxy-based connection layer. Is this an oversight, or a
deliberate change related to 0.9.1?

Sigurd

Search Discussions

  • Aaron Westendorf at Aug 30, 2010 at 6:38 pm
    Some time ago we had a conversation with the team and learned that the
    redirect was removed in the 0.9.1 spec. I suspect that the
    corresponding known_hosts data may have also been removed. A quick
    check of the protocol docs seem to confirm this.

    -Aaron

    2010/8/30 Sigurd H?gsbro <sigurd.hogsbro at museeka.com>:
    Hi,

    Since the upgrade to 2.0.0 I notice that the known_hosts property is
    no longer getting populated by the hostname of the server to which
    I've connected. This was quite useful to us for diagnostics since we
    use a proxy-based connection layer. Is this an oversight, or a
    deliberate change related to 0.9.1?

    Sigurd
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-discuss at lists.rabbitmq.com
    https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss


    --
    Aaron Westendorf
    Senior Software Engineer
    Agora Games
    359 Broadway
    Troy, NY 12180
    Phone: 518.268.1000
    aaron at agoragames.com
    www.agoragames.com
  • Sigurd Høgsbro at Aug 31, 2010 at 1:16 am
    Thanks Aaron, I was afraid that's what would be coming back ;-)

    Sigurd
    On Aug 30, 8:38?pm, Aaron Westendorf wrote:
    Some time ago we had a conversation with the team and learned that the
    redirect was removed in the 0.9.1 spec. ?I suspect that the
    corresponding known_hosts data may have also been removed. ?A quick
    check of the protocol docs seem to confirm this.

    -Aaron

    2010/8/30 Sigurd H?gsbro <sigurd.hogs... at museeka.com>:
    Hi,
    Since the upgrade to 2.0.0 I notice that the known_hosts property is
    no longer getting populated by the hostname of the server to which
    I've connected. This was quite useful to us for diagnostics since we
    use a proxy-based connection layer. Is this an oversight, or a
    deliberate change related to 0.9.1?
    Sigurd
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-disc... at lists.rabbitmq.com
    https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
    --
    Aaron Westendorf
    Senior Software Engineer
    Agora Games
    359 Broadway
    Troy, NY 12180
    Phone: 518.268.1000
    aa... at agoragames.comwww.agoragames.com
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-disc... at lists.rabbitmq.comhttps://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
  • Alexis Richardson at Aug 31, 2010 at 12:51 pm
    Sigurd

    Do you think you will be able to develop a workaround?

    alexis


    2010/8/31 Sigurd H?gsbro <sigurd.hogsbro at museeka.com>:
    Thanks Aaron, I was afraid that's what would be coming back ;-)

    Sigurd
    On Aug 30, 8:38?pm, Aaron Westendorf wrote:
    Some time ago we had a conversation with the team and learned that the
    redirect was removed in the 0.9.1 spec. ?I suspect that the
    corresponding known_hosts data may have also been removed. ?A quick
    check of the protocol docs seem to confirm this.

    -Aaron

    2010/8/30 Sigurd H?gsbro <sigurd.hogs... at museeka.com>:
    Hi,
    Since the upgrade to 2.0.0 I notice that the known_hosts property is
    no longer getting populated by the hostname of the server to which
    I've connected. This was quite useful to us for diagnostics since we
    use a proxy-based connection layer. Is this an oversight, or a
    deliberate change related to 0.9.1?
    Sigurd
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-disc... at lists.rabbitmq.com
    https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
    --
    Aaron Westendorf
    Senior Software Engineer
    Agora Games
    359 Broadway
    Troy, NY 12180
    Phone: 518.268.1000
    aa... at agoragames.comwww.agoragames.com
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-disc... at lists.rabbitmq.comhttps://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-discuss at lists.rabbitmq.com
    https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
  • Sigurd Høgsbro at Aug 31, 2010 at 1:57 pm
    Hi Alexis,

    As it is mainly for diagnostics we'll deal without it.

    Regards,

    Sigurd
    On 31 Aug 2010, at 13:51, Alexis Richardson wrote:

    Sigurd

    Do you think you will be able to develop a workaround?

    alexis


    2010/8/31 Sigurd H?gsbro <sigurd.hogsbro at museeka.com>:
    Thanks Aaron, I was afraid that's what would be coming back ;-)

    Sigurd
    On Aug 30, 8:38 pm, Aaron Westendorf wrote:
    Some time ago we had a conversation with the team and learned that the
    redirect was removed in the 0.9.1 spec. I suspect that the
    corresponding known_hosts data may have also been removed. A quick
    check of the protocol docs seem to confirm this.

    -Aaron

    2010/8/30 Sigurd H?gsbro <sigurd.hogs... at museeka.com>:
    Hi,
    Since the upgrade to 2.0.0 I notice that the known_hosts property is
    no longer getting populated by the hostname of the server to which
    I've connected. This was quite useful to us for diagnostics since we
    use a proxy-based connection layer. Is this an oversight, or a
    deliberate change related to 0.9.1?
    Sigurd
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-disc... at lists.rabbitmq.com
    https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
    --
    Aaron Westendorf
    Senior Software Engineer
    Agora Games
    359 Broadway
    Troy, NY 12180
    Phone: 518.268.1000
    aa... at agoragames.comwww.agoragames.com
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-disc... at lists.rabbitmq.comhttps://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-discuss at lists.rabbitmq.com
    https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
  • Sigurd Høgsbro at Sep 1, 2010 at 9:42 am
    Alexis,

    A follow-on thought: could you add a key with the server hostname to the server_properties dict sent back from the server on connection.start? That would solve our diagnostics needs (and maybe Aaron's too) without breaking your conformance to the spec.

    Regards,

    Sigurd
    On 31 Aug 2010, at 14:57, Sigurd H?gsbro wrote:

    Hi Alexis,

    As it is mainly for diagnostics we'll deal without it.

    Regards,

    Sigurd
    On 31 Aug 2010, at 13:51, Alexis Richardson wrote:

    Sigurd

    Do you think you will be able to develop a workaround?

    alexis


    2010/8/31 Sigurd H?gsbro <sigurd.hogsbro at museeka.com>:
    Thanks Aaron, I was afraid that's what would be coming back ;-)

    Sigurd
    On Aug 30, 8:38 pm, Aaron Westendorf wrote:
    Some time ago we had a conversation with the team and learned that the
    redirect was removed in the 0.9.1 spec. I suspect that the
    corresponding known_hosts data may have also been removed. A quick
    check of the protocol docs seem to confirm this.

    -Aaron

    2010/8/30 Sigurd H?gsbro <sigurd.hogs... at museeka.com>:
    Hi,
    Since the upgrade to 2.0.0 I notice that the known_hosts property is
    no longer getting populated by the hostname of the server to which
    I've connected. This was quite useful to us for diagnostics since we
    use a proxy-based connection layer. Is this an oversight, or a
    deliberate change related to 0.9.1?
    Sigurd
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-disc... at lists.rabbitmq.com
    https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
    --
    Aaron Westendorf
    Senior Software Engineer
    Agora Games
    359 Broadway
    Troy, NY 12180
    Phone: 518.268.1000
    aa... at agoragames.comwww.agoragames.com
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-disc... at lists.rabbitmq.comhttps://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-discuss at lists.rabbitmq.com
    https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
  • Michael Bridgen at Sep 1, 2010 at 10:21 am
    Sigurd,
    A follow-on thought: could you add a key with the server hostname to
    the server_properties dict sent back from the server on
    connection.start? That would solve our diagnostics needs (and maybe
    Aaron's too) without breaking your conformance to the spec.
    The server properties is more or less a free-form map, so yes, it's
    certainly possible a hostname (or more likely, a node name including the
    hostname) could go there.

    There may be concerns around leaking infrastructural information that
    way, but probably not strong ones.


    Michael
    Regards,

    Sigurd
    On 31 Aug 2010, at 14:57, Sigurd H?gsbro wrote:

    Hi Alexis,

    As it is mainly for diagnostics we'll deal without it.

    Regards,

    Sigurd

    On 31 Aug 2010, at 13:51, Alexis Richardson <alexis at rabbitmq.com>
    wrote:
    Sigurd

    Do you think you will be able to develop a workaround?

    alexis


    2010/8/31 Sigurd H?gsbro <sigurd.hogsbro at museeka.com>:
    Thanks Aaron, I was afraid that's what would be coming back ;-)



    Sigurd

    On Aug 30, 8:38 pm, Aaron Westendorf <aa... at agoragames.com>
    wrote:
    Some time ago we had a conversation with the team and learned
    that the redirect was removed in the 0.9.1 spec. I suspect
    that the corresponding known_hosts data may have also been
    removed. A quick check of the protocol docs seem to confirm
    this.

    -Aaron

    2010/8/30 Sigurd H?gsbro <sigurd.hogs... at museeka.com>:
    Hi, Since the upgrade to 2.0.0 I notice that the
    known_hosts property is no longer getting populated by the
    hostname of the server to which I've connected. This was
    quite useful to us for diagnostics since we use a
    proxy-based connection layer. Is this an oversight, or a
    deliberate change related to 0.9.1? Sigurd
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-disc... at lists.rabbitmq.com
    https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
    -- Aaron Westendorf Senior Software Engineer Agora Games 359
    Broadway Troy, NY 12180 Phone: 518.268.1000
    aa... at agoragames.comwww.agoragames.com
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-disc... at lists.rabbitmq.comhttps://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
    _______________________________________________
    rabbitmq-discuss mailing list
    rabbitmq-discuss at lists.rabbitmq.com
    https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss

    _______________________________________________ rabbitmq-discuss
    mailing list rabbitmq-discuss at lists.rabbitmq.com
    https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
  • Matthias Radestock at Sep 1, 2010 at 10:26 am
    Sigurd,
    On 01/09/10 11:21, Michael Bridgen wrote:
    A follow-on thought: could you add a key with the server hostname to
    the server_properties dict sent back from the server on
    connection.start? That would solve our diagnostics needs (and maybe
    Aaron's too) without breaking your conformance to the spec.
    The server properties is more or less a free-form map, so yes, it's
    certainly possible a hostname (or more likely, a node name including the
    hostname) could go there.

    There may be concerns around leaking infrastructural information that
    way, but probably not strong ones.
    I am reluctant to add ad-hoc properties without a sound technical
    rationale. And yes the information leakage *is* a concern.

    Can you explain what you are trying to do?

    connection.redirect and the known_hosts field got removed because they
    a) duplicate what load balancers do much better, and b) don't work in
    many network setups since they rely on the clients to be able to resolve
    servers by their internal names.


    Regards,

    Matthias.
  • Sigurd Høgsbro at Sep 1, 2010 at 12:22 pm
    Hi Matthias,

    We use the hostname for follow-up diagnosis after identifying application
    failures. In such scenarios it is useful to know which of the RabbitMQ
    servers behind the (transparent) load-balancer to investigate.

    Fair point on the information leakage, but I'm not sure why you're concerned
    about the server hostname being exposed as it used to be exposed until
    recently. We could also use a name/token specified in the server
    configuration file if you prefer not to expose DNS names.

    Otherwise we'll have to figure a way get this information from the
    load-balancers.

    Sigurd

    2010/9/1 Matthias Radestock <matthias at rabbitmq.com>
    Sigurd,

    On 01/09/10 11:21, Michael Bridgen wrote:

    A follow-on thought: could you add a key with the server hostname to
    the server_properties dict sent back from the server on
    connection.start? That would solve our diagnostics needs (and maybe
    Aaron's too) without breaking your conformance to the spec.
    The server properties is more or less a free-form map, so yes, it's
    certainly possible a hostname (or more likely, a node name including the
    hostname) could go there.

    There may be concerns around leaking infrastructural information that
    way, but probably not strong ones.
    I am reluctant to add ad-hoc properties without a sound technical
    rationale. And yes the information leakage *is* a concern.

    Can you explain what you are trying to do?

    connection.redirect and the known_hosts field got removed because they a)
    duplicate what load balancers do much better, and b) don't work in many
    network setups since they rely on the clients to be able to resolve servers
    by their internal names.


    Regards,

    Matthias.
    -------------- next part --------------
    An HTML attachment was scrubbed...
    URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20100901/c560df7d/attachment.htm>
  • Matthias Radestock at Sep 1, 2010 at 12:55 pm
    Sigurd,
    On 01/09/10 13:22, Sigurd H?gsbro wrote:
    Fair point on the information leakage, but I'm not sure why you're
    concerned about the server hostname being exposed as it used to be
    exposed until recently.
    We all learn from our mistakes ;)
    We could also use a name/token specified in the server configuration
    file if you prefer not to expose DNS names.
    Now *there* is an idea. Being able to extend/augment the
    server_properties as part of the server configuration, much like our
    client APIs allow you to do with the client properties. See, e.g.
    http://www.rabbitmq.com/releases/rabbitmq-java-client/v2.0.0/rabbitmq-java-client-javadoc-2.0.0/com/rabbitmq/client/ConnectionFactory.html#setClientProperties(java.util.Map)

    I have filed a bug for that.


    Regards,

    Matthias.
  • Sigurd Høgsbro at Sep 1, 2010 at 1:02 pm
    Matthias,

    Excellent; configurable server properties work for me.

    Sigurd

    2010/9/1 Matthias Radestock <matthias at rabbitmq.com>
    Sigurd,

    On 01/09/10 13:22, Sigurd H?gsbro wrote:

    Fair point on the information leakage, but I'm not sure why you're
    concerned about the server hostname being exposed as it used to be
    exposed until recently.
    We all learn from our mistakes ;)


    We could also use a name/token specified in the server configuration
    file if you prefer not to expose DNS names.
    Now *there* is an idea. Being able to extend/augment the server_properties
    as part of the server configuration, much like our client APIs allow you to
    do with the client properties. See, e.g.
    http://www.rabbitmq.com/releases/rabbitmq-java-client/v2.0.0/rabbitmq-java-client-javadoc-2.0.0/com/rabbitmq/client/ConnectionFactory.html#setClientProperties(java.util.Map)

    I have filed a bug for that.


    Regards,

    Matthias.
    -------------- next part --------------
    An HTML attachment was scrubbed...
    URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20100901/c39a18c1/attachment.htm>

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
grouprabbitmq-discuss @
categoriesrabbitmq
postedAug 30, '10 at 5:55p
activeSep 1, '10 at 1:02p
posts11
users5
websiterabbitmq.com
irc#rabbitmq

People

Translate

site design / logo © 2022 Grokbase