On 8.6.2016 05:15, Fraser Tweedale wrote:
On Tue, Jun 07, 2016 at 03:42:22PM +1000, Fraser Tweedale wrote:
On Wed, Jun 01, 2016 at 02:51:04PM +1000, Fraser Tweedale wrote:
This patchset implements the 'ca' plugin for creating and managing
lightweight sub-CAs, and updates the 'caacl' plugin and
'cert-request' command to support multiple CAs.
A brief overview of the patches:
'ca' plugin, associated schema changes and container objects,
Dogtag REST API wrapper
Add CA entry for the IPA CA on install/upgrade
Update 'caacl' plugin with CA support (including enforcement)
Update ra.request_certificate() to support specifying target CA
Add '--ca' option to 'cert-request' command
Add '--issuer' option to 'cert-find' command
These patches depend on other pending patches:
0051, 0052, 0053, 0054, 0055, 0056
Signing key replication depends on unmerged Dogtag patches. Builds
of Dogtag with the required patches, and of FreeIPA with all
completed sub-CAs work, should be available from my COPR soon:
Some parts of the design are not implemented in the current
- local parent CA (ipaca object) references
- sub-CA certificate renewal
- 'cert-show' command '--ca=NAME' option
- certmonger support for specifying CA
- revocation of deleted CAs
I look forward to your reviews!
Rebased and updated patches attached.
- add required attributes for issuer DN and subject DN
- prevent rename of IPA CA
- when adding IPA CA entry, contact Dogtag to learn authority id,
issuer DN and subject DN
- add 'read_ca' method to Dogtag interface
- tighten ACIs to prevent modification of ipacaid attribute
Updated patch 0064-3; adds --issuer option to cert-show and --ca
option to cert-show and cert-find.
1) On upgrade, why is the lightweight CA container created twice - once
in 41-subca.update, once using ensure_entry() call? It should be done
2) In ca_del, every CA specified in args should be deleted, not just
the first one.
3) Do not use NonFatalError, issue a warning instead:
4) Can it actually happen that ca_show does not return ipacaid? I guess
not, so you should be able to remove the check altogether and don't
bother with the warning.
Patch 0060-0062: LGTM
Could you please define the CA param as follows:
doc=_("CA to use"),
This is for consitency with framework-generated parent key params, which
unfortunately we cannot leverage in cert_request currently.
1) See my comment for patch 0063, it applies here as well.
2) The --issuer option should not be included in cert_show - show
commands are supposed to retrieve an object given primary key(s), and
the primary key of CA objects is just their cn.
3) In find commands, the options form a filter, so instead of raising
MutuallyExclusiveError in cert-find, return an empty result, as with any
other unmatched filter.
Manage your subscription for the Freeipa-devel mailing list:
Contribute to FreeIPA: http://www.freeipa.org/page/Contribute/Code