<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=ISO-8859-1">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<meta http-equiv="content-type" content="text/html;
charset=ISO-8859-1">
<meta http-equiv="content-type" content="text/html;
charset=ISO-8859-1">
Enabling heat on an otherwise default devstack will increase the
keystone catalog so that the resulting v3 token header exceeds
eventlet.wsgi's limit [1] of 8192 bytes.<br>
<br>
This is currently preventing Heat from being enabled in gating
[2][3], and others have hit this limit too [4].<br>
<br>
The catalog-optional blueprint [5] might help, but it would probably
need to become the default fairly soon, since the current token size
is not far under the limit.<br>
<br>
One possible (unverified) fix would be to patch every project's
wsgi.py to raise eventlet.wsgi.
<meta http-equiv="content-type" content="text/html;
charset=ISO-8859-1">
MAX_HEADER_LINE to provide some headroom (32768? that leaves another
32768 for other headers)<br>
<br>
Is there any objection to this? I can start with nova and keystone
to see if it fixes the failing tempest tests.<br>
<br>
[1] <a
href="https://github.com/eventlet/eventlet/blob/master/eventlet/wsgi.py#L19">https://github.com/eventlet/eventlet/blob/master/eventlet/wsgi.py#L19</a><br>
<meta http-equiv="content-type" content="text/html;
charset=ISO-8859-1">
[2] <a href="https://bugs.launchpad.net/keystone/+bug/1190149">https://bugs.launchpad.net/keystone/+bug/1190149</a><br>
[3] <a href="https://review.openstack.org/#/c/32002/">https://review.openstack.org/#/c/32002/</a><br>
[4] <a href="https://bugs.launchpad.net/keystone/+bug/1186177">https://bugs.launchpad.net/keystone/+bug/1186177</a><br>
[5] <a
href="https://blueprints.launchpad.net/keystone/+spec/catalog-optional">https://blueprints.launchpad.net/keystone/+spec/catalog-optional</a><br>
</body>
</html>