Requests is an HTTP library, written in Python, for human beings. This library adds optional Kerberos/GSSAPI authentication support and supports mutual authentication. Basic GET usage:
>>> import requests
>>> from requests_kerberos import HTTPKerberosAuth
>>> r = requests.get("http://example.org", auth=HTTPKerberosAuth())
...
The entire requests.api
should be supported.
In order to use this library, there must already be a Kerberos Ticket-Granting
Ticket(TGT) cached in a Kerberos credential cache. Whether a TGT is available
can be easily determined by running the klist
command. If no TGT is
available, then it first must be obtained by running the kinit
command, or
pointing the $KRB5CCNAME to a credential cache with a valid TGT.
In short, the library will handle the "negotiations" of Kerberos authentication, but ensuring that an initial TGT is available and valid is the responsibility of the user.
Client authentication failures will be communicated to the caller by returning the 401 response. A 401 response may also come from an expired Ticket-Granting Ticket.
By default, HTTPKerberosAuth
will require mutual authentication from the
server, and if a server emits a non-error response which cannot be
authenticated, a requests_kerberos.errors.MutualAuthenticationError
will
be raised. If a server emits an error which cannot be authenticated, it will
be returned to the user but with its contents and headers stripped. If the
response content is more important than the need for mutual auth on errors,
(eg, for certain WinRM calls) the stripping behavior can be suppressed by
setting sanitize_mutual_error_response=False
:
>>> import requests
>>> from requests_kerberos import HTTPKerberosAuth, REQUIRED
>>> kerberos_auth = HTTPKerberosAuth(mutual_authentication=REQUIRED, sanitize_mutual_error_response=False)
>>> r = requests.get("https://windows.example.org/wsman", auth=kerberos_auth)
...
If you'd prefer to not require mutual authentication, you can set your
preference when constructing your HTTPKerberosAuth
object:
>>> import requests
>>> from requests_kerberos import HTTPKerberosAuth, OPTIONAL
>>> kerberos_auth = HTTPKerberosAuth(mutual_authentication=OPTIONAL)
>>> r = requests.get("http://example.org", auth=kerberos_auth)
...
This will cause requests_kerberos
to attempt mutual authentication if the
server advertises that it supports it, and cause a failure if authentication
fails, but not if the server does not support it at all.
While we don't recommend it, if you'd prefer to never attempt mutual authentication, you can do that as well:
>>> import requests
>>> from requests_kerberos import HTTPKerberosAuth, DISABLED
>>> kerberos_auth = HTTPKerberosAuth(mutual_authentication=DISABLED)
>>> r = requests.get("http://example.org", auth=kerberos_auth)
...
HTTPKerberosAuth
can be forced to preemptively initiate the Kerberos
GSS exchange and present a Kerberos ticket on the initial request (and all
subsequent). By default, authentication only occurs after a
401 Unauthorized
response containing a Kerberos or Negotiate challenge
is received from the origin server. This can cause mutual authentication
failures for hosts that use a persistent connection (eg, Windows/WinRM), as
no Kerberos challenges are sent after the initial auth handshake. This
behavior can be altered by setting force_preemptive=True
:
>>> import requests
>>> from requests_kerberos import HTTPKerberosAuth, REQUIRED
>>> kerberos_auth = HTTPKerberosAuth(mutual_authentication=REQUIRED, force_preemptive=True)
>>> r = requests.get("https://windows.example.org/wsman", auth=kerberos_auth)
...
If communicating with a host whose DNS name doesn't match its
kerberos hostname (eg, behind a content switch or load balancer),
the hostname used for the Kerberos GSS exchange can be overridden by
setting the hostname_override
arg:
>>> import requests
>>> from requests_kerberos import HTTPKerberosAuth, REQUIRED
>>> kerberos_auth = HTTPKerberosAuth(hostname_override="internalhost.local")
>>> r = requests.get("https://externalhost.example.org/", auth=kerberos_auth)
...
HTTPKerberosAuth
normally uses the default principal (ie, the user for
whom you last ran kinit
or kswitch
, or an SSO credential if
applicable). However, an explicit principal can be specified, which will
cause Kerberos to look for a matching credential cache for the named user.
This feature depends on OS support for collection-type credential caches,
as well as working principal support in PyKerberos (it is broken in many
builds). An explicit principal can be specified with the principal
arg:
>>> import requests
>>> from requests_kerberos import HTTPKerberosAuth, REQUIRED
>>> kerberos_auth = HTTPKerberosAuth(principal="user@REALM")
>>> r = requests.get("http://example.org", auth=kerberos_auth)
...
On Windows, WinKerberos is used instead of PyKerberos. WinKerberos allows the
use of arbitrary principals instead of a credential cache. Passwords can be
specified by following the form user@realm:password
for principal
.
requests_kerberos
supports credential delegation (GSS_C_DELEG_FLAG
).
To enable delegation of credentials to a server that requests delegation, pass
delegate=True
to HTTPKerberosAuth
:
>>> import requests
>>> from requests_kerberos import HTTPKerberosAuth
>>> r = requests.get("http://example.org", auth=HTTPKerberosAuth(delegate=True))
...
Be careful to only allow delegation to servers you trust as they will be able to impersonate you using the delegated credentials.
This library makes extensive use of Python's logging facilities.
Log messages are logged to the requests_kerberos
and
requests_kerberos.kerberos_
named loggers.
If you are having difficulty we suggest you configure logging. Issues with the underlying kerberos libraries will be made apparent. Additionally, copious debug information is made available which may assist in troubleshooting if you increase your log level all the way up to debug.