Wednesday, October 9, 2019

PGP Signing with SignServer


This blog post covers PGP signing support implemented in recent versions of SignServer

In a previous blog post, we addressed Code Signing of Windows binaries (Authenticode) and gave some background on why Code Signing is important for secure software distribution.

We are now pleased to introduce code signing with PGP, commonly used for Open Source software projects and packaging of software for Linux environments in general.

OpenPGP Signing was first implemented in SignServer Enterprise 5.1 and will be available in SignServer Community 5.2, already available for download as a beta release.

SignServer Installation

To test the PGP support, download and install SignServer Enterprise 5.1 or SignServer Community 5.2.0.Beta1 or later.

For installation instructions, refer to SignServer Installation.

Setting up an OpenPGP Signer

When SignServer is up and running, access the SignServer Administration Web to start setting up the workers.

The Administration Web is available in SignServer Community as if 5.2.0.Beta1.

Step 1: Set up Crypto Worker

If you don’t already have a crypto worker configured, then set one up using, for example, the sample keystore:
  1. Select the AdminWeb Workers tab, and click Add.
  2. Click From Template, select keystore-crypto.properties in the list, and click Next.
  3. In the configuration text view, change the value for WORKERGENID1.KEYSTOREPATH so that the path corresponds to your SignServer installation, for example: WORKERGENID1.KEYSTOREPATH=/home/username/signserver/res/test/dss10/dss10_keystore.p12. 
  4. Click Apply.
Remember the name of the crypto worker (for example, CryptoTokenP12) as you will need it in the next step when setting up the OpenPGP Signer.

Step 2: Set up OpenPGP Signer

To set up the new OpenPGP signer, do the following:
  1.  Select the SignServer AdminWeb Workers tab, and click Add to add a new worker.
  2. Choose the method From Template.
  3. Select openpgpsigner.properties in the Load from Template list and click Next.
  4. Change the sample configuration properties as needed, for example:
    • Update the CRYPTOTOKEN property with the name of your crypto worker (for example, CryptoTokenP12).
    • Update DEFAULTKEY to an existing key (or do this in a later step).
  1. Click Apply to load the configuration and list the worker in the All Workers list.
  2. Select the added worker in the list to open the Worker page.
  3. Check if the Worker status is Offline and if there are any errors listed. The "No key available for purpose" message means that the DEFAULTKEY property does not point to an existing key in the crypto token. In that case, either update the DEFAULTKEY property to point to an existing key or do the following to generate a new key to use with this signer:
    • Click Renew key and specify the following:
      • Set a Key Algorithm, for example "RSA".
      • Set a Key Specification, for example the key length "2048" (for RSA).
      • Update the New Key Alias to the name of DEFAULTKEY property (typically change to the same value as the Old Key Alias).
    • Click Generate.
    • Select the worker in the list and confirm that the Worker status is Active and without errors listed. If not, confirm that the DEFAULTKEY property is correct and check in the Crypto Token tab of the crypto worker that a key with the specified name exists.

Step 3: Add User ID to Public Key

Follow the steps below to add User ID / Certification for the OpenPGP public key using the Generate CSR option.
  1. Select the AdminWeb Workers tab.
  2. Select the OpenPGP worker and click Generate CSR.
  3. Specify a Signature Algorithm, for example "SHA256withRSA"
  4. Specify DN as the wanted User Id, for example "Markus (Code Signing) <markus@primekey.se>".
  5. Click Generate, and then click Download.
  6. Open the downloaded file using any text editor and copy its content.
  7. Select the worker and click the Configuration tab.
  8. For the PGPPUBLICKEY property, click Edit.
  9. Paste the public key content in the Value field, and click Submit.
  10. Click Status Summary and confirm that fields like PGP Key ID and PGP Public key are listed. Also, note that the User ID is listed. 

Step 4: Sign

The following example shows how to sign using the SignServer Public Web. You can test signing using any of the SignServer client interfaces.
  1. Click Client Web.
  2. Under File Upload, specify the Worker name used, for example, OpenPGPSigner.
  3. Select the file to create a detached signature for, for example, release.zip.
  4. Click Submit and store the resulting signature file, for example, release.zip.asc.

Step 5: Verify Signature

The following example shows how to verify the signature using the OpenPGP tool GnuPG. It should be possible to use any OpenPGP tool to verify the signature.

Run the following to verify the signature using GnuPG:
$ gpg --verify release.zip.asc release.zip

If needed, first import the public key to GnuPG before verifying the signature in the third step:
  1. Store the public key (i.e. from PGPPUBLICKEY property) as signer001-pub.asc.
  2. Import the key to GnuPG:
    $ gpg --import signer001-pub.asc

More Information

See also the OpenPGP Signer documentation and if you have any questions or comments don't hesitate to use our discussion forums.

Cheers,
the PrimeKey SignServer Team


Thursday, June 27, 2019

EJBCA 7.2.0 - CT improvements & Extended REST API

Summer is here and as promised, so is EJBCA 7.2.0! Highlighted news are performance improvements to Certificate Transparency and additional functionality added to the REST API.

Persistent Storage of Certificate Transparency SCT Responses

Persistent caching of Certificate Transparency SCTs (Signed Certificate Timestamps), in the form of a database-backed storage, has been added in addition to the existing in-memory caching. This reduces the number of requests to the CT log server and increases the performance in the following ways:

  •  The database-backed storage will be used after a restart when the in-memory cache is empty.
  • The in-memory storage has a limit of 100 000 certificates by default, and will only keep the SCTs for the most recently requested certificates. The database-backed storage has no such limit and will be used for SCTs for less frequently requested certificates.
  • The database-backed storage will store partial results for a certificate, allowing EJBCA to retry a submission efficiently at a later point.

Additionally, the default configuration was changed to rate-limit connections to logs that are down or return error codes. This reduces the load on both log servers and
EJBCA. For example, if a CT log rate-limits EJBCA, then EJBCA will back off for 1 second by default.

Crypto Token and CA Management REST API

The EJBCA REST API has so far been limited to Certificate Management operations. We've now extended the REST API, adding resources for CA administration as well. This allows simpler remote integration and management as an option to the GUI. New endpoints support Crypto Token and CA Management including:
  • CA activation and deactivation.
  • Crypto Token activation and deactivation.
  • Key generation and removal.
New REST end points in Swagger-UI
















Expect more news from us this autumn!

Cheers,
the PrimeKey EJBCA Team

Tuesday, June 25, 2019

EJBCA ♥ YubiKey

With the keygen tag in its final death throes, time has come to move on to new and better ways of managing keys on tokens. We here at PrimeKey are big fans of our friends at Yubico, so here is a neat little guide of how to get up and running with using your YubiKey with EJBCA.

Disclaimer: This blog post is in no way sponsored nor endorsed by Yubico, though they were quite kind and provided us with a couple of tokens to play around with. We just honestly like them. 

Prerequisites

To get going, you're going to need to have the following installed on your workstation:

Creating a key pair on your YubiKey


1. Start up the YubiKey Manager
2. Under Applications, pick PIV and then Configure Certificates

3. Under the Authentication tab, click Generate to create a new key pair on the token. This will take you through a guide of creating a key pair by picking an algorithm, key size and setting the Common Name for your token. 
4. Make sure that you specify that you was a Certificate Signing Request (CSR)
5. This will result in a CSR that you can use to enroll the key pair to EJBCA. 

Enrolling the YubiKey to EJBCA

Enrolling the newly created key pair is done just like with any other enrollment.

1. Go to the EJBCA RA UI
2. Click on Enroll, choose the appropriate certificate type and sub-type and choose Generated by User, which will prompt you to upload your CSR from the previous step.
3. Fill in any other pertinent information, then choose Download PEM

Importing the Certificate to the YubiKey 

1. Open the YubiKey Manager, and again choose Applications, pick PIV and then Configure Certificates
2. Click Import and pick the new newly generated certificate.
3. Congratulations - your YubiKey is now up and running, but we still need to configure the browser to play along. 

Configuring FireFox to use YubiKey

1. Open Firefox and enter about:preferences in the address bar
2. Under Privacy and Security click on Security Devices 
3. Click on Load to install OpenSC's PKCS#11 Driver
4. Name the module and then locate the opensc-pkcs11.so (or similar) library
5. The YubiKey will now be shown as a security module

Configuring Access Rights in EJBCA 

This is done using Roles as with any user EJBCA administrator, but here are the exact steps:

1. In the EJBCA CA UI, pick Roles and either create a new Role or add the new administrator to an existing Role
2. To add the administrator, click on Members, pick the appropriate CA and enter the identifying the information for the certificate, preferably the serial number
3. An easy way to find the serial number is to view the certificate in OpenSSL using the command:
$ openssl x509 -in alanwidget.pem -text -noout
The serial number can be copied, converted from hex to decimal using a converter and then used in EJBCA.

4. Finally, click on Access Rules and set the required rules for your administrator
5. The next time you start a new session, your YubiKey will be offered as an option

Thursday, May 16, 2019

A bit about us...

We seldom speak much about the team behind EJBCA or about PrimeKey Solutions on a whole on this blog, so for those interested I'd like to write a bit about us and our culture for those of you who might be interested.

The Evolution of a Company

PrimeKey Solutions was founded way back when in 2002 by a small group of developers who saw the need for an Open Source CA solution, which was innovatively named EJBCA. While we're great at many things, we do admit that naming products is maybe not one of them - I do bid you remember that this was the tail end of the IT-era; it could have been way worse. 
PrimeKey's logo ca 2005
FOSS was an important concept from the very start - both from a practical viewpoint (don't trust anything that you can't verify), but also ideological. PrimeKey was founded on the principal that cryptography (and by extension PKI) should be widely available, so was from the very start founded on FOSS principles, and PrimeKey have ever sought to make use of open source, as well as contributing to the projects that we use (such as BouncyCastle) both in code and monetarily. 
PrimeKey's logo ca 2014
PrimeKey has since then grown and evolved, from being a tiny one-room-everybody-does-everything-company to what it is today: a steadily growing, +100 person company with offices in three countries, specialized staff and specific roles going forward. 
Our logo since 2017

The Evolution of a Team

I joined PrimeKey back in 2010, back in the heydays of the entire company sharing the same volume of air during a working day. Except for our then CEO and a couple of other staff, the bunch of us (<10 people) mostly filled the same roles to varying degrees: developer, professional services, IT, support, QA, documentation, tech sales - there was little delimitation and even less structure. You did what was needed and learned the skills required to overcome the next task at hand. We were happy in those days, though we were disorganized. 

I was originally hired as a developer, but back in those rousing days the prospect of travel appealed to me as well. Between writing (and refactoring) mounds of code, working with customers took me from Ankara to Antwerpen, from Brussels to Baku. My first major project was working with our upcoming Common Criteria certification back in 2011, which taught me the joys of writing months worth of pointless documentation and speaking to auditors.

What I'm getting at was that we existed in mostly a state of unmitigated pandemonium; working on customer requirements as they arose, our own fancies and interests, putting out wild bush fires and occasionally lighting them. At a tech conference in 2012 I described our work style as "Controlled Chaos", which very much makes me cringe today. 

This finally started changing round about 2016 or so, when we were faced with a few issues:
  • We were evolving as a company, moving more towards corporate customers. A couple of bad experiences of letting customers specify implementations, implementing those specs to the letter and then discovering that the customer had asked for the wrong thing showed that we needed to involve ourselves more in specification work. 
  • While the free-form organization was great for allowing the removal of tech debt on a minor scale, non-critical refactoring that would have required major effort were difficult to organize. 
  • Our roadmap was basically just the JIRA backlog, meaning that release dates were largely hypothetical concepts. 
  • Long release cycles (3-4 months) coupled with a lack of issue slicing and cycle breaks (such as sprints) lead to monolithic commits waiting for months on review, occasionally turning out long after the fact to be unsatisfactory or incomplete. 
We direly needed to change. I won't go into the whole process and evolution of how we got to where we are today (though I might write that as a blog post at a later date), I'd love to describe how the EJBCA development team works at present. 

The Team(s)

The goals of our teams are to be as self organizing as possible, wherein each has their own integrated Scrum Master in order to organize and synchronize the teams around common goals. With QA as integral parts of each team, we've started moving away from testing as a waterfall feature and instead testing proactively (and most importantly, just testing). 

The teams each maintain their own culture and rules internally, and sync one day a week in either a general tech meeting to discuss ongoing tasks, hinderances or questions, and have a common retrospective at the end of each sprint in order to evaluate what went well and what went poorly. 

Our Roadmap and Development Cycle 

My role as Product Owner is primarily to maintain a vision and a roadmap in order to help the teams focus on the most important requirements for the moments, to give guidance and answer implementation questions that the teams can't answer themselves. Thus my main avenue of communication with the teams is maintaining and constantly updating the roadmap, which allows them to on their own prioritize and sort the backlog (with some input from myself). Versions are set for features and fixes only as they enter the near horizon (except for features with fixed dates such as customer implementations and specification implementations), which sends less false signals to customers in regards to completion dates, and gives a truer image of what can be expected in each release.

The teams today follow three week development sprints, at the end of which all issues need to be resolved, reviewed and closed. Issues are sliced down to a three days or less if possible. This cuts down on both issue size and time-to-review, letting us catch design errors earlier, before they propagate. It also eliminates the large degree of release date uncertainty due to the large grey blob of review time (and resulting fixes) always invariably plaguing the end of each release cycle. Besides the retrospective mentioned above, sprints often end with an internal sprint demo to show what's been accomplished, and a release demo post release to show off for the rest of the company what's new. 

The Result

In the last couple of years we've taken great leaps forward in terms in terms of reliability, predictability and above all quality. While we still have a long way to go, we are extremely pleased with the results so far.

The Evolution of an Office 

The Stockholm office has changed addresses four times in nearly 20 years. The gang started off sitting in what amounted to a shoebox, before moving to slightly bigger digs in 2009, which is about the time I came into the picture.

The new office was slightly larger than the old one, but the entire team still shared a single room (with an office for our CEO and VP Sales), happily all breathing the same air for the entire working day. 

My desk back in 2010, with era-appropriate camera filter and still relevant stacks of DVD blanks. 
We shared an office back then with two other companies, with constant bickering over which coffee machine to lease (bad coffee leads to bad code - you can quote me on that). We were on top of a children's osteopath, so on clear days the workday was hellishly audioscaped by the little moppets' wailings up through the ventilation. Good times. 

With time we came to outgrow that office, first shifting out one of our office-mates before getting outshifted ourselves. We'd grown to expand over two rooms, and there was as little piece and quiet as there was space to move around the tightly packed desks. Upon finding new space in 2016, I sat down and wrote a manifesto of office design, arguing that the current and continued practices of open offices was provably based on faulty theories on workspace design, citing the esteemed Joel Spolsky, among other sources. I ranted! I raved! I built barricades out of hutches and obsolete cryptographic digests. Join me! I called. Let us throw off the yoke of office plan oppression, seize the means of production and be allowed to work without external interruptions! 

But one colleague heeded my call. We were given our own room in the new office and told to shut up. We were happy. 
The author exploring his new digs. New feats of concentration were attained in this small room. 
Come this year (2018) we had once more outgrown the previous office. PrimeKey's success in the last few years led to huge growth for the company, and we finally expanded into having proper teams handling support, integration and sales - by extension allowing us to hire more developers and QA staff. Management located a new place only a stone's throw away from the existing one, this time moving us into a complete floor of a proper office building, complete with proper server room in the basement, a view and space for luxuries like a gaming room. 
Our office since 2018
A gaming room with a foosball table, the true sign of prosperity
Upon planning the move our CEO Magnus very graciously (and wisely) asked a selection of employees our thoughts on the new office and ideas for an office plan, so for the developers we suggested a scheme that lies more in line with both our own philosophies and with what is considered modern practice. We actively rejected any suggestions for open office plans (or its blasphemous mutated step child, hot desking), but instead suggested that we put up walls or roof-to-floor-dividers, and space them out to create reasonably sound proof work areas containing 4-6 desks, which we consider the ideal team size. We felt that this would have the following advantages:
  • Humans are social creatures, and do well from interacting with others. Development work (particularly in cross functional teams) requires the ability to hold spontaneous meetings and interactions. For teams to form bonds they need to interact, tell in-jokes, form their own culture. 
  • Development work generally requires large degrees of concentration due to the mental effort of visualizing complex solutions, and unwanted distractions (others' phones, distant conversations, etc) lead to stress, anxiety and generally lower productivity. 
  • As humans we tend to nest in order to create a sense of belonging and familiarity in our surroundings, spreading around us items both useful (that extra phone charger) and decorative. We tend to do so in small groups as well, so this scheme would allow each team to be able to decorate their own spaces as they see fit. 
In addition, the additions of walls and partitions would allow each team to have a dedicated whiteboard to use while collaborating on features, a large TV monitor to statically display information (such as CI machine status, roadmaps or JIRA boards). In the end I have to say, management delivered beautifully:
Team Alice's work area

The home of Team Bob
The dwellings of the SignServer Team
While it may not look like much, these spaces serve us perfectly in allowing each team to concentrate on problems and issues, allowing collaboration and social interaction while letting those needing to bite down on a problem concentrate while avoiding the angst-inducing sight of cubicle farms. 

All in all, were constantly evolving here at PrimeKey - analyzing and reevaluating what makes us better as individuals, as teams and as a company.

We're already doing what we love, and now we're a huge leap forward in how much we love doing it. 

Cheers,
Mike Agrenius Kushner
Product Owner EJBCA

Monday, April 29, 2019

EJBCA 7.1.0 - Partitioned CRLs!

Spring has finally arrived in Stockholm, following the traditional seasons of Winter, False Spring, Second Winter, the Spring of Deceit and the final cold snap of I-Just-Changed-My-Tires. The melting snows bring with them many gifts, besides the beer forgotten on the balcony last November, among them EJBCA 7.1

Partitioned CRLs

Long and enduringly requested, EJBCA 7.1 is now capable of producing partitioned CRLs. Activated under the CA configuration, the number of partitions per CRL is dynamically configurable, allowing new partitions to be added as the CRL grows, and assignment to older partitions to be suspended in order to allow for future growth. CDP partition assignment is random in order to allow for even distribution of certificates, and partition definition can be looked up in the CDP extension as defined in RFC5280.
For those of you not wishing to use partitioned CRLs life will mostly move on as usual while for those of you applying partitioned CRLs to existing installations you will retain a legacy CRL for pre-existing certificates (as the CDP can't be changed retroactively) while newly issued certificates will be issued to partitions.

Deprecation and Removal of Hard Token Support

In an effort to relieve ourselves of maintaining little-used features we have chosen in this release to deprecate and remove support of hard tokens, after analyzing that it has little to no use among PrimeKey customers. Naturally this will have no impact on existing installations, but we have provided scripts for those of you wishing to remove the relevant tables from the database. See the upgrade notes for more details.

VA and RA Specific Distributions

As a response to market interest, we've enhanced our build process and modularization in order to produce VA and RA specific builds of EJBCA, each capable of acting in their specific roles but not as a CA. This allows PrimeKey to offer a more dynamic model for Appliance and Cloud users who would like to add RA and VA instances to their PKIs but find it prohibitive to pay for the full fee for the complete distribution. The standard CA distribution still retains the full VA and RA capabilities as before. If you're interested in finding out more, please contact sales@primekey.com

EJBCA 6.15.2 CE Available on Docker Hub

As some of you already know, as part of our ongoing containerization project we've added a docker container to Docker Hub, built on a sneak-peek of the coming release of EJBCA 6.15.2 Community Edition. 
If you're interested in moving your PKI towards containerization, please go ahead and have a look, and feel free to give us any feedback! 

Cheers,
Mike Agrenius Kushner
Product Owner, EJBCA

Monday, March 4, 2019

EJBCA 7.0.1 - PSD2 and SN Entropy

Hot on the heels of EJBCA 7.0, we'd like to present the release of EJBCA 7.0.1 - implementing a ton of neat functionality that didn't make the cut for the main release. On top of the list of most commonly requested features is PSD2 support, but please read on to find all the reasons to upgrade to EJBCA 7.0.1!

Full PSD2 Support

EJBCA 7.0.1 provides full support for the Payment Services Directive as defined by EU Directive 2015/2366. PSD2 allows eIDAS Trusted Certificate Providers to issue PSD2 QWAC certificates to third party FinTech companies, which in turn gives them access to financial APIs hosted by European banks. To enable PSD2 in your instance of EJBCA, scroll down to the QC Statements extension of your certificate profile and enable the PSD2 option. 
 This will enable PSD2 fields in the RA UI during enrollment.
If you'd like to read more about PSD2, we've written blog post about it on PrimeKey's blog and right here on our own development blog.  

Domain Blacklist Validator

As a request from some of our CABF-customers, we've implemented a Domain Blacklist Validator. The new Validator takes a list of partial and complete domain names, and can be configured to either block them outright (if run during the data phase) or cause an approval action to be triggered in the final approval step (if approvals are activated). 
All of the approving RA administrators in the final approval step will be shown the following warning before the approval passes:

dnsName SAN can be Automatically Populated by the CN

We've added a setting to End Entity Profiles to allow the dnsName Subject Alternative Name field in a certificate to be filled in by the Common Name (CN) value in the Subject DN.

Configurable SN Entropy, Default Value Raised to 20 Octets

CA/B Forum requires the use of 64 bit entropy when generating serial numbers (see CABF Ballot 164). Due to only positive values being valid serial numbers, 8 octets will only result in 63 bit entropy as the most-significant-bit will always be 0, hence we recommend larger sizes than 8 octets. Previously this was set using the property ca.serialnumberoctetsize in cesecore.properties, which has now been dropped and the value is instead set directly in the CA.
Possible values may range between 4 and 20 octets, and the default for all new CAs is 20 while upgraded CA's will retain whatever value was set in ca.serialnumberoctetsize, or 8 if none was set. 

Downloadable CSRs

In EJBCA 7.0.1 we've started storing CSRs along with the associated certificate (instead of only the last submitted CSR as it was earlier), so you now have access to download and review all CSRs submitted and processed in the past.

URL Metadata Type Added to Approvals

Upon popular request, we've added a URL metadata type to the partitioned approval profiles. It allows the approving RA administrator to enter a URL while performing the approval, e.g pointing to a file upload at an external location.
Upon later review of the approval, it will show up as a hyperlink:

Experimental: Configuration Checker

Lastly, we're trying out an experimental new feature in EJBCA 7.0.1, the Configuration Checker. It displays an (incomplete) list of common configuration issues on the front page. 
If you'd like to try it out, it can be activated in its own tab under the System Configuration:

Roadmap Update

Common Criteria

Our common criteria process is ongoing - the Security Target (ST) is now complete and has been sent for evaluation. Preliminary date for a certified version of EJBCA is still projected to be at the end of this summer. 

Appliance Release

EJBCA 7.0.1 will be available on Appliance 3.3.0, due at the end of March/beginning of April.

Up Next

The teams are rearing to go to work on EJBCA 7.1. Main features are going to be Partitioned CRLs, multi-value RDN support and a couple of surprises. See you then!

Cheers,
Mike Agrenius Kushner
Product Owner, EJBCA