<div dir="ltr"><div><div><div><div><div>Hi,<br><br></div>I manage the IdP software much like I manage other software at my organization. I use svn and break it out into a few separate products (shibboleth, tomcat, shibbolth configuration).<br><br></div>The "shibboleth" product is for housing the actual IdP code. It gets touched on upgrades and hopefully that's it.<br><br>The "Tomcat" product is just the Tomcat files.<br><br></div>The "shibboleth configuration" product is all the metadata, configuration, etc.<br><br></div>We have a custom release management software to make packages and release them, but basically it's just a Web UI that makes a tarball from svn and gives it a build number. Clicking to release just means it will scp relevant tarball(s) to a server and unpack them.<br><br>On release (or you can do with symlinks or the like) it will build connections between the products -- e.g. Tomcat needs to know about the idp.war file.<br><br>I also have things organized to have shib, apache, and tomcat logs write to a data directory separate from where they normally would be written (and I have an instance of Splunk where the prod log data goes too).<br><br></div><div>That's the basics of how I manage my instance Shibboleth IdP. In terms of servers I have:<br><ul><li>1 dev VM</li><li>2 test VMs, load balanced</li><li>2 prod VMs, load balanced</li></ul><p>Having the servers behind a load balancer means I can restart one at a time without impacting end users (though unless I tell my systems administrators to manually point the load balancer at one server the IdP may be down for ~30 seconds for some users before it realizes one of the servers is down).<br></p><p>I hope that helps! Let me know if you have any questions.<br></p><p>Ian<br></p></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Tue, Feb 3, 2015 at 1:56 PM, Benjamin Cherian <span dir="ltr"><<a href="mailto:benjamin.cherian@villanova.edu" target="_blank">benjamin.cherian@villanova.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">For those that have been using IdP for a while, what solution(s) are you<br>
using to manage instances of IdP and changed made to the configuration. Do<br>
you use a VCS like Git or SVN, or another tool like Chef or Ansible? What<br>
does your setup look like how does your deployment process look like?<br>
<br>
Currently we're looking to setup a load balanced solution with 2 IdP<br>
servers in both dev and production. We want a solid solution for managing<br>
when we add new RPs, change attribute filters, and make other<br>
configuration changes.<br>
<br>
Thanks,<br>
Ben Cherian<br>
<span class="HOEnZb"><font color="#888888"><br>
--<br>
To unsubscribe from this list send an email to <a href="mailto:users-unsubscribe@shibboleth.net">users-unsubscribe@shibboleth.net</a><br>
</font></span></blockquote></div><br><br clear="all"><br>-- <br><div class="gmail_signature">Ian Rifkin '04, MS '09<br>Software Systems Manager<br>Library and Technology Services (LTS)<br>Brandeis University<br><br><a href="http://go.brandeis.edu/u:irifkin" target="_blank">http://go.brandeis.edu/u:irifkin</a></div>
</div>