Unless I'm not understanding something there isn't much change to process for TinkerPop. You spot a problem in IronBank and either submit a PR to fix or create a JIRA for someone else to fix if you don't have the ability to do it yourself. It gets merged and presumably you can get your image into IronBank.
What I don't understand is how you consume that fix. Let's say IronBank had a problem in 3.6.2. TinkerPop fixes it. Now what? Do you have to wait until we release 3.6.3 to get that into IronBank and hope you don't hit more problems? Is there a way to test an image ahead of a TinkerPop release? I guess I'm trying to confirm that there really isn't any change to anything TinkerPop does in the normal process of reviewing pull request and fixing security problems because it seems pretty normal/simple otherwise. No one at TinkerPop can release anything to IronBank so this is purely a third-party maintained convenience which you are handling for the wider community. Is that a fair depiction of what were looking at here? So, if there's nothing out of the ordinary to do from TinkerPop's side then I guess you should feel free to let the PRs/JIRAs fly (or maybe just call attention to existing ones if dependabot is already on it). Perhaps you should ensure you comment on these that they are "IronBank" related so that they perhaps get some visibility that way. On Wed, Feb 22, 2023 at 9:48 AM Jim Foscue <[email protected]> wrote: > I would like to bring this discussion back up. > > Has anyone given this any more thought? My team is willing to help but we > have not participated in the development of any of tinkerpop so it would be > a huge effort on our part to get involved in the development. We are > developers but not our main language is C#. > > Thanks > Jim Foscue > > On Wed, Dec 21, 2022 at 5:40 AM Stephen Mallette <[email protected]> > wrote: > > > From what I'm gathering, this doesn't sound as if it changes much of how > > things operate here or the release process. It sounds as though Jim > simply > > needs to submit dependency related PRs to TinkerPop. They would be > reviewed > > and merged in similar fashion to any PR that arrived. As these would be > > security related PRs, they probably wouldn't generate objections/concerns > > and would merge quickly. > > > > There is still something I don't quite understand. 3.6.1 scanned by > > IronBank shows problems. Jim submits PRs to TinkerPop to fix those > issues. > > What version ultimately lands in IronBank? IronBank has to wait for a > 3.6.2 > > as 3.6.1 can't pass their scanners? Can IronBank scanners work from > > SNAPSHOT? Ideally, IronBank would be working on 3.6.2-SNAPSHOT right now > as > > release is arriving within the next few weeks. Then you could have a > clean, > > pre-tested artifact in 3.6.2 to push into IronBank when it releases. This > > way IronBank would align with DockerHub and the rest of the release. > > Putting some kind of patched version of 3.6.1 in IronBank seems like it > > could cause confusion imo. > > > > On Tue, Dec 20, 2022 at 12:51 PM Jim Foscue <[email protected] > > > > wrote: > > > > > I believe most of the critical/high vulnerabilities are a result of > > > outdated dependencies. > > > > > > Another thing to consider is that to participate as an active developer > > on > > > Ironbank the user must have a DoD CAC which allows them access to the > > > Ironbank repository and the vulnerabilities. I'm not sure if anyone > here > > > has one but I do. I was able to get the process started with the 3.6.1 > > > version and there are a number of critical and highs. Some of the > > examples > > > are: > > > Apache Shiro before 1.5.3, when using Apache Shiro with Spring dynamic > > > controllers, a specially crafted request may cause an authentication > > > bypass. > > > Neo4j through 3.4.18 (with the shell server enabled) exposes an RMI > > service > > > that arbitrarily deserializes Java objects, e.g., through > > > setSessionVariable. An attacker can abuse this for remote code > execution > > > because there are dependencies with exploitable gadget chains. > > > HttpObjectDecoder.java in Netty before 4.1.44 allows a Content-Length > > > header to be accompanied by a second Content-Length header, or by a > > > Transfer-Encoding header. > > > > > > Looks like the Apache Shiro library is one of the problem spots. > > > > > > I do agree with your assessment that including a scanner in the normal > > > build process would help a lot. > > > > > > Jim > > > > > > > > > On Tue, Dec 20, 2022 at 3:08 AM Florian Hockmann < > [email protected] > > > > > > wrote: > > > > > > > It would be good to know what kind of vulnerabilities Ironbank can > > find. > > > > Most of such scanners I know simply use a database of known > > > vulnerabilities > > > > that include the libraries and their versions that are affected. The > > > > Log4Shell vulnerability from last year is a good example that could > be > > > > found by such a scanner. > > > > > > > > In general I think that we should ideally try to include such a > scanner > > > > into our usual dev process so we can find those vulnerabilities early > > and > > > > then fix them. Fixing them will probably most of the time just mean > > > > updating a dependency. > > > > If we just add Ironbank as an additional release channel behind our > > usual > > > > release process, then we just get notified about vulnerabilities > after > > we > > > > just have released a version with these vulnerabilities. I think that > > > would > > > > only lead to additional pain for us as we would have to fix them and > > then > > > > start another release process to perform a release without the > > > > vulnerability. > > > > > > > > GitHub already offers such a vulnerability scanner that works > together > > > > with Dependabot. This means that Dependabot will create a PR to > update > > a > > > > dependency if that contains a known vulnerability. These security > > related > > > > PRs will be marked as such, but in a way that is only visible to > > > > maintainers of the repo. So, we can prioritize these security related > > PRs > > > > over the usual Dependabot PRs that sometimes overwhelm us due to the > > high > > > > number of dependencies we have. > > > > This improves the security of TinkerPop in general and I think that > it > > > > also improves our dev process as we currently get tickets for such > > > security > > > > problems manually created by users who probably already execute such > > > > scanners on our JARs / Docker images (two recent examples: > > > > TINKERPOP-2843[1] and TINKERPOP-2826[2]) and then they are usually > also > > > > addressed by someone who manually updates the dependency. > > > > If the Ironbank scanner works in a similar way, then this would also > > make > > > > it a lot easier to publish our images to the Ironbank registry as > that > > > > should ideally not uncover any new vulnerabilities since they have > > > already > > > > been found and fixed by our own process (like GitHub security > scanning > > + > > > > Dependabot). > > > > > > > > By the way, I think this is the documentation of Ironbank: > > > > https://p1.dso.mil/products/iron-bank > > > > And here is a list of available Docker images: > > > https://repo1.dso.mil/dsop > > > > Unfortunately, the repos seem to be private so we can't see much, but > > the > > > > list at least gives an idea of what kind of projects are already > > > available > > > > there. > > > > > > > > [1]: https://issues.apache.org/jira/browse/TINKERPOP-2843 > > > > [2]: https://issues.apache.org/jira/browse/TINKERPOP-2826 > > > > > > > > -----Ursprüngliche Nachricht----- > > > > Von: Jim Foscue <[email protected]> > > > > Gesendet: Montag, 19. Dezember 2022 17:10 > > > > An: [email protected] > > > > Betreff: Re: [DISCUSS] > > > > > > > > Yes, Ironbank is an approved DoD dockerhub. There is no requirement > > for > > > > the releases on Ironbank to be in sync with the releases in > > > > Apache/Tinkerpop. The major requirement for Ironbank is for there > not > > to > > > > be any critical/high vulnerabilities in the docker image. If there > are > > > > vulnerabilities that "pop" up they should be addressed quickly. > > > > > > > > My team can help with that process but it would be hard for us to fix > > the > > > > vulnerabilities without knowing the code. > > > > > > > > Jim > > > > > > > > On Mon, Dec 19, 2022 at 5:36 AM Stephen Mallette < > [email protected] > > > > > > > wrote: > > > > > > > > > Hi Jim, thanks for bringing this discussion here. > > > > > > > > > > Generally speaking, it sounds like a reasonable idea to open > > TinkerPop > > > > > opportunities to the DoD environment. Would it be fair to call > > > > > IronBank a DoD approved DockerHub? Would the idea be that TinkerPop > > > > > would release to IronBank in the same cadence as is done with the > > > > > standard release schedule that publishes to DockerHub? > > > > > > > > > > If so, I think a concern would be that supporting IronBank creates > a > > > > > hurdle to releasing that may be high and thus delay releases that > > > > > might otherwise be acceptable outside of that space. I don't know > if > > > > > it makes sense, but perhaps releasing to IronBank should be > decoupled > > > > from standard releases? > > > > > Like 3.5.5 would go out the door and then a separate process would > go > > > > > under way to harden 3.5.5 to meet IronBank needs? Of course, then > the > > > > > 3.5.5 in IronBank isn't necessarily the 3.5.5 in DockerHub so that > > > > > could be problematic. Do you happen to know how other major open > > > > > source projects do this? > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Wed, Dec 14, 2022 at 4:51 PM Jim Foscue > > > > > <[email protected]> > > > > > wrote: > > > > > > > > > > > Ironbank discussion > > > > > > > > > > > > I would like to suggest making tinkerpop/gremlin available on the > > > > > > DoD Ironbank docker registry (https://ironbank.dso.mil). > > > > > > > > > > > > Ironbank is a DoD docker registry that maintains publicly > available > > > > > docker > > > > > > container images that have been "hardened' and given the ok to > use > > > > > > on DoD contracts. The process of hardening an image involves > > > > > > running the docker container image through a vulnerability > scanning > > > > > > process. Any critical > > > > > and > > > > > > high vulnerabilities have to be addressed before getting the > docker > > > > > > container image approved for DoD use. > > > > > > > > > > > > My team has been using the publicly available tinkerpop/graph on > a > > > > > > DoD contract that is a prototype. But to continue using it in a > > > > > > more > > > > > official > > > > > > process we have to have it approved by Ironbank. > > > > > > > > > > > > Therefore, I am trying to get the tinkerpop developer community > to > > > > > > get onboard to push this to Ironbank so that we and other DoD > > > > > > contracts can > > > > > use > > > > > > it. > > > > > > > > > > > > We have access to the Ironbank system (which requires a US > > > > > > government > > > > > CAC) > > > > > > and are willing to help guide the process. > > > > > > > > > > > > Please consider this request and provide any feedback to me at > your > > > > > > convience. > > > > > > > > > > > > Regards, > > > > > > > > > > > > Jim Foscue > > > > > > JDM Solutions > > > > > > > > > > > > > > > > > > > > > > > > -- > > > > > > *Jim Foscue*, Software Engineer > > > > > > *JDM Solutions* > > > > > > 256.694.9387 > > > > > > [email protected] <[email protected]> > > > > > > > > > > > > -- > > > > > > *This e-mail and any attachements are for the intended > recipient(s) > > > > > > only and may contain information proprietary or private to JDM > > > > Solutions, LLC. > > > > > > If you are not the intended recipient **please delete without > > > > > > copying and kindly advise the sender by e-mail of the mistake in > > > > > > delivery. This > > > > > email > > > > > > may be personal in communication from the sender and as such does > > > > > > not represent the views of the company.* > > > > > > > > > > > > > > > > > > > -- > > > > *This e-mail and any attachements are for the intended recipient(s) > > only > > > > and may contain information proprietary or private to JDM Solutions, > > LLC. > > > > If you are not the intended recipient **please delete without copying > > and > > > > kindly advise the sender by e-mail of the mistake in delivery. This > > > email > > > > may be personal in communication from the sender and as such does not > > > > represent the views of the company.* > > > > > > > > > > > > > > -- > > > *This e-mail and any attachements are for the intended recipient(s) > only > > > and may contain information proprietary or private to JDM Solutions, > LLC. > > > If you are not the intended recipient **please delete without copying > and > > > kindly advise the sender by e-mail of the mistake in delivery. This > > email > > > may be personal in communication from the sender and as such does not > > > represent the views of the company.* > > > > > > > -- > *This e-mail and any attachments are for the intended recipient(s) only > and > may contain information proprietary or private to JDM Solutions, LLC. If > you are not the intended recipient **please delete without copying and > kindly advise the sender by e-mail of the mistake in delivery. This email > may be personal in communication from the sender and as such does not > represent the views of the company.* >
