Hi! Reviving this thread, thank you, Eron, for starting this and for the preparation of the FLIP. I am sharing some thoughts below, and some input based on what has changed with FLIP-6 and the evolution of queryable state.
Best, Stephan ------------------------------- *Internal vs. External Connectivity* That is a very helpful distinction, let's build on that. - I would suggest to treat eventually all communication coming potentially from users as external, meaning Client-to-Dispatcher, Client-to-JobManager (trigger savepoint, change parallelism, ...), Web UI, Queryable State. - That leaves communication that is only between JobManager/TaskManager/ ResourceManager/Dispatcher/HistoryServer as internal. - I am somewhat operating under the assumption that all external communication will eventually be HTTP/REST. That works best with many setups and is the basis for using service proxies that handle authentication/authorization. In Flink 1.5 and future versions, we have the following update there: - Akka is now strictly internal connectivity, the client (except legacy client) do not use it any more. - The Blob Server will move to purely internal connectivity in Flink 1.6, where a POST of a job to the Dispatcher has the jars and the JobGraph. That is important for Kubernetes setups, where exposing the BlobServer and querying the blob port causes quite some friction. - Treating queryable state as "internal connectivity" is fine for now. We should treat it as "external" connectivity in the future if we move it to HTTP/REST. *Internal Connectivity and SSL Mutual Authentication* Simply activating SSL mutual authentication for the internal communication is a really low hanging fruit. Activating client authentication for Akka, network stack Netty (and Blob Server/Client in Flink 1.6) should require no change in the configurations with respect to Flink 1.4. All processes are, with respect to internal communication, simultaneously server and client endpoints. Because of that, they already need KeyStore and TrustStore files for SSL handshakes, where the TrustStore needs to trust the KeyStore Certificate. I personally favor the suggestion made to have a script that generates a self-signed certificate and adds it to "conf" and updates the configuration. That should be picked up by the Yarn and Mesos clients anyways. *External Connectivity* There is a huge surface area and I think we need to give users a way to plug in their own tools. >From what I see (and after some discussions with Patrick and Gary) I think it makes sense to look at proxies in a broad way, similar to the approach Eron outlined. The basic approach could be like that: - Everything goes through HTTPS, so the proxy can work with HTTP headers. - The proxy handles authentication and possibly authorization. The proxy adds some header, for example a user name, a group id, an authorization token. - Flink can configure an implementation of an 'authorizer' or validator on the headers to decide whether the request is valid. - Example 1: The proxy does authentication and adds the user name / group as a header. The the Flink-side authorizer simply checks whether the name is in the config (simple ACL-style) scheme. - Example 2: The proxy adds an JSON Web Token and the authorizer validates that token. For secure connections between the Proxy and the Flink Endpoint I would follow Eron's suggestion, to use separate KeyStores and TrustStores than for internal communication. For Yarn and Mesos, I would like to see if we could handle those again as a special case of the proxies above: - DCOS Admin Router forwards the user authentication token, so that could be another authorizer implementation. - In YARN we could see if can implement the IP filter via such an authorizer. *Hostname Verification* I am not sure if the suggestion in the FLIP means to not use hostname verification at all, or to always use it (no configuration flag). Eron, can you clarify what you mean there? For Kubernetes, it is very hard to work with certificates and have hostname verification on. If we assume internal communication works strictly with a shared secret certificate and with client authentication, does hostname verification actually still add security in that particular setup? My understanding was that hostname verification is important to not have some valid certificate presented, but the one bound to the server you want to talk to. If we have anyways one trusted certificate only, isn't that already implied? On the other hand, it is still possible (and potentially valuable) for users in standalone mode to use keystores and truststores from a PKI, in which case there may still be an argument in favor of hostname verification. On Wed, Sep 27, 2017 at 6:46 AM, Eron Wright <eronwri...@gmail.com> wrote: > Hi folks, I'm happy to share with you a draft of a FLIP for service > authorization. As I mentioned at the top of this thread, the goal is to > protect a deployed Flink cluster/session from unauthorized use. In the > doc, I propose the use of SSL client authentication for internal > communication, plus YARN/Kubernetes/Mesos-specific functionality to achieve > single sign-on for the web interface/API. > > FLIP-? - Service Authorization > <https://docs.google.com/document/d/13IRPb2GdL842rIzMgEn0ibO > QHNku6W8aMf1p7gCPJjg/edit?usp=sharing> > > Feel free to make comments and to give feedback on whether this feature > would be useful to you. > > Thanks! > Eron > > > On Thu, Aug 3, 2017 at 11:11 AM, Eron Wright <eronwri...@gmail.com> wrote: > > > Till, with (c) are you suggesting that we'd use Akka 2.3 for Scala 2.10 > > and Akka 2.4+ for Scala 2.11+? Sounds reasonable but I don't know how > > feasible it is. I will say I'm optimistic because a) Akka 2.4 is said > to > > be binary compatible, and b) the Flakka fork appears to be subsumed by > 2.4. > > > > Let us then take (c) as the tentative plan. > > > > I agree the community should discuss dropping Scala 2.10 but I don't want > > to drive that conversation. > > > > Thanks > > > > > > On Thu, Aug 3, 2017 at 6:24 AM, Ufuk Celebi <u...@apache.org> wrote: > > > >> I haven't followed this discussion in detail nor am I familiar with > >> the service authorization topic or Flakka, but a) sounds like a lot of > >> maintenance work to me. > >> > >> If possible I would go with c) and maybe start a discussion about > >> dropping Scala 2.10 support to check whether that is a viable option > >> or not. > >> > >> – Ufuk > >> > >> > >> On Thu, Aug 3, 2017 at 1:59 PM, Till Rohrmann <trohrm...@apache.org> > >> wrote: > >> > Alternatively there would also be an option > >> > > >> > c) only support mutual auth for Akka 2.4+ if the backport is > >> unrealistic to > >> > do > >> > > >> > But this of course would break security for Scala 2.10. On the other > >> hand > >> > people are already using Flink without this feature. > >> > > >> > Cheers, > >> > Till > >> > > >> > On Wed, Aug 2, 2017 at 7:21 PM, Eron Wright <eronwri...@gmail.com> > >> wrote: > >> > > >> >> Thanks Till and Aljoscha for the feedback. > >> >> > >> >> Seems there are two ways to proceed here, if we accept mutual SSL as > >> the > >> >> basis. > >> >> > >> >> a) Backport mutual-auth support from Akka 2.4 to Flakka. > >> >> b) Drop support for Scala 2.10 (FLINK-?), move to Akka 2.4 > >> (FLINK-3662). > >> >> > >> >> Let's assume (a) for now. > >> >> > >> >> > >> >> > >> >> On Tue, Aug 1, 2017 at 3:05 PM, Till Rohrmann <trohrm...@apache.org> > >> >> wrote: > >> >> > >> >> > Dropping Java 7 alone is not enough to move to Akka 2.4+. For that > we > >> >> need > >> >> > at least Scala 2.11. > >> >> > > >> >> > Cheers, > >> >> > Till > >> >> > > >> >> > On Tue, Aug 1, 2017 at 4:22 PM, Aljoscha Krettek < > >> aljos...@apache.org> > >> >> > wrote: > >> >> > > >> >> > > Hi Eron, > >> >> > > > >> >> > > I think after Dropping support for Java 7 we will move to Akka > >> 2.4+, so > >> >> > we > >> >> > > should be good there. I think quite some users should find a > (more) > >> >> > secure > >> >> > > Flink interesting. > >> >> > > > >> >> > > Best, > >> >> > > Aljoscha > >> >> > > > On 24. Jul 2017, at 03:11, Eron Wright <eronwri...@gmail.com> > >> wrote: > >> >> > > > > >> >> > > > Hello, now might be a good time to revisit an important > >> enhancement > >> >> to > >> >> > > > Flink security, so-called service authorization. This means > the > >> >> > > hardening > >> >> > > > of a Flink cluster against unauthorized use with some sort of > >> >> > > > authentication and authorization scheme. Today, Flink relies > >> >> entirely > >> >> > > on > >> >> > > > network isolation to protect itself from unauthorized job > >> submission > >> >> > and > >> >> > > > control, and to protect the secrets contained within a Flink > >> cluster. > >> >> > > > This is a problem in multi-user environments like > YARN/Mesos/K8. > >> >> > > > > >> >> > > > Last fall, an effort was made to implement service > authorization > >> but > >> >> > the > >> >> > > PR > >> >> > > > was ultimately rejected. The idea was to add a simple secret > >> key to > >> >> > all > >> >> > > > network communication between the client, JM, and TM. Akka > >> itself > >> >> has > >> >> > > > such a feature which formed the basis of the solution. There > are > >> >> > > usability > >> >> > > > challenges with this solution, including a dependency on SSL. > >> >> > > > > >> >> > > > Since then, the situation has evolved somewhat, and the use of > >> SSL > >> >> > mutual > >> >> > > > authentication is more viable. Mutual auth is supported in > Akka > >> >> > 2.4.12+ > >> >> > > > (or could be backported to Flakka). My proposal is: > >> >> > > > > >> >> > > > 1. Upgrade Akka or backport the functionality to Flakka (see > >> commit > >> >> > > > 5d03902c5ec3212cd28f26c9b3ef7c3b628b9451). > >> >> > > > 2. Implement SSL on any endpoint that doesn't yet support it > >> (e.g. > >> >> > > > queryable state). > >> >> > > > 3. Enable mutual auth in Akka and implement it on non-Akka > >> endpoints. > >> >> > > > 4. Implement a simple authorization layer that accepts any > >> >> > authenticated > >> >> > > > connection. > >> >> > > > 5. (stretch) generate and store a certificate automatically in > >> YARN > >> >> > mode. > >> >> > > > 6. (stretch) Develop an alternate authentication method for the > >> Web > >> >> UI. > >> >> > > > > >> >> > > > Are folks interested in this capability? Thoughts on the use > of > >> SSL > >> >> > > mutual > >> >> > > > auth versus something else? Thanks! > >> >> > > > > >> >> > > > -Eron > >> >> > > > >> >> > > > >> >> > > >> >> > >> > > > > >
