[
https://issues.apache.org/jira/browse/MESOS-2902?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14619857#comment-14619857
]
Cody Maloney commented on MESOS-2902:
-------------------------------------
In DCOS we do all Mesos config via environment variables (Allows better mixing
and matching in various environemnts). We ship the same mesos-master systemd
unit to every cluster, and then we change the configuration by swapping out
environment variable files (See Systemd's {{EnvironmentFile}} directive).
Inside an {{EnvironmentFile}} we can't run arbitrary scripts. It is
structurally in-feasible to change the mesos-master systemd unit per cluster to
include the 'Set the IP by running this script' only in cases where we want to
do that. There may also cases where Mesos exits and we restart it, and it would
refuse to start because it has a different IP (mesos slave might checkpoint it,
although I'd have to double check).
The IP to use is a per-host thing, so I can't ship a generic config file to
every host in the cluster which just sets {{LIBPROCESS_IP}} in an
{{EnvironmentFile}}.
Writing a wrapper script which sets {{LIBPROCESS_IP}} and then does an {{exec
mesos-master}} is feasible, although it obfuscates what is happening, and if
someone we ship DCOS to has been hand-editing the script for their environment
and gets the environment variable a little bit wrong, things will error really
badly (We've had a number of customers with mesos figuring out that the host's
IP is 127.0.0.1).
As far as the hostname stuff: In general we need Mesos not to do anything with
hostnames in a number of our environments because they are unreliable, esp. as
a means for figuring out "what address should I talk on".
> Enable Mesos to use arbitrary script / module to figure out IP, HOSTNAME
> ------------------------------------------------------------------------
>
> Key: MESOS-2902
> URL: https://issues.apache.org/jira/browse/MESOS-2902
> Project: Mesos
> Issue Type: Improvement
> Components: master, modules, slave
> Reporter: Cody Maloney
> Assignee: Marco Massenzio
> Priority: Critical
> Labels: mesosphere
>
> Currently Mesos tries to guess the IP, HOSTNAME by doing a reverse DNS
> lookup. This doesn't work on a lot of clouds as we want things like public
> IPs (which aren't the default DNS), there aren't FQDN names (Azure), or the
> correct way to figure it out is to call some cloud-specific endpoint.
> If Mesos / Libprocess could load a mesos-module (Or run a script) which is
> provided per-cloud, we can figure out perfectly the IP / Hostname for the
> given environment. It also means we can ship one identical set of files to
> all hosts in a given provider which doesn't happen to have the DNS scheme +
> hostnames that libprocess/Mesos expects. Currently we have to generate
> host-specific config files which Mesos uses to guess.
> The host-specific files break / fall apart if machines change IP / hostname
> without being reinstalled.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
