Inedo Community Forums Forums
    • Recent
    • Tags
    • Popular
    • Login
    1. Home
    2. sai.pabbareddy
    3. Posts

    Welcome to the Inedo Forums! Check out the Forums Guide for help getting started.

    If you are experiencing any issues with the forum software, please visit the Contact Form on our website and let us know!

    S Offline
    • Profile
    • Following 0
    • Followers 0
    • Topics 10
    • Posts 28
    • Groups 0

    Posts

    Recent
    • Missing default GPL rule

      Fresh install doesn't ship with any default GPL/copyleft compliance rule

      Quick question on License Compliance from our Enterprise trial evaluation. Our understanding from your documentation was that a fresh ProGet install ships with some kind of default policy flagging GPL/copyleft licenses for review. On our fresh Enterprise trial instance, that wasn't the case — a package we tested (MySql.Data 8.4.0, correctly detected as GPL-2.0-only) was Compliant by default, with no Specified License Rule in place at all until we added one ourselves.

      Once we added a rule marking GPL-2.0-only as Noncompliant, enforcement worked exactly as expected — the same real 400 Bad Request block we see for vulnerability-based quarantine applied identically.

      Is a default GPL/copyleft rule meant to be pre-configured out of the box, and we're missing a setup step, or is that something every install needs to configure explicitly regardless of edition? Want to make sure we don't have a false sense of security assuming default license enforcement exists when it doesn't.

      Thanks,
      Sai

      posted in Support
      S
      sai.pabbareddy
    • Restart-dependent config caching

      Permission and directory config changes don't take effect until a service restart

      We've run into this twice during our Enterprise trial evaluation and want to check if it's expected behavior.

      Case 1 — RBAC: We removed two permission grants from the Anonymous user/group ("Administer globally" and "View & Download Packages globally") via pgutil security permissions remove. pgutil security permissions list immediately confirmed the grants were gone from the database. Despite that, requests with fabricated/wrong credentials (which should now fall through with no permissions at all) continued to succeed at both package downloads and admin-level actions (creating a feed) immediately after the change. Only after we ran docker restart proget did the fix actually take effect — re-running the identical tests afterward, everything correctly failed with 401/403.

      Case 2 — Directory config: Separately, while configuring an LDAP user directory, we changed the "User Search Base" field and saved it. The LDAP server's own connection log showed ProGet still querying with the old (incorrect) search base on the next login attempt. Again, only a docker restart proget made the new value actually take effect.

      Is this expected — i.e., does permission/directory configuration get cached in memory and only reloaded on service restart? If so:

      • Is there a lighter-weight way to force that cache to refresh in a production deployment, short of a full service restart?
      • Is this documented anywhere, so we can build it into our own change-management process?

      Thanks,
      Sai

      posted in Support
      S
      sai.pabbareddy
    • OpenLDAP directory: authentication bind uses empty DN instead of resolved user

      Continuing our Enterprise trial evaluation — this time on the "OpenLDAP/Generic LDAP" user directory type. We stood up a real OpenLDAP server (Docker, osixia/openldap), created a real user, and confirmed that user could bind successfully outside of ProGet (ldapwhoami succeeded). We were never able to get ProGet to authenticate that same user, and traced it to three separate issues, each confirmed against the LDAP server's own connection log:

      1. Auto-derived search base is wrong when left blank

      With the "User Search Base" field empty, ProGet derived a search base by treating our "Host" field's IP address as if it were a dotted domain name:
      Host: 172.17.0.1
      Derived base: dc=172,dc=17,dc=0,dc=1
      Every search failed with LDAP error 32 ("no such object") until we set "User Search Base" explicitly to our real directory root (dc=progetpoc,dc=local).

      1. The %s username placeholder in the "Users:" filter isn't substituted

      The "Users:" field ships pre-filled by ProGet itself with (&(objectClass=inetOrgPerson)(uid=%s)). With a correct search base in place, the query actually sent to the LDAP server was (&(objectClass=inetOrgPerson)(?uid=)) — the %s collapsed to a literal ?, and the submitted username never made it into the filter. This matched zero users regardless of who was authenticating.

      1. The credential-verification bind uses an empty DN instead of the resolved user

      To isolate #2, we temporarily hardcoded the filter to (uid=ldapuser1). The search then correctly found the user (nentries=1). But the very next operation — presumably meant to verify the submitted password by binding as that user's DN — was instead BIND dn="" method=128, which succeeded trivially (an anonymous bind checks no password). ProGet then failed the overall request with an internal "No Such Object" error, surfaced to the client as a 500.

      Full log excerpt:
      SRCH base="dc=progetpoc,dc=local" filter="(&(objectClass=inetOrgPerson)(uid=ldapuser1))"
      SEARCH RESULT tag=101 err=0 nentries=1
      BIND dn="" method=128
      RESULT tag=97 err=0
      ProGet: "No Such Object" -> HTTP 500 to the client

      Issue #3 is the one that actually blocks things — even a hand-corrected configuration that resolves the right user can't complete authentication. Is this a known issue, and is there a recommended workaround, or is this expected to be fixed in an upcoming build?

      Thanks,
      Sai

      posted in Support
      S
      sai.pabbareddy
    • Does Active Directory support require domain-joining, even with "Domain controller host" set?

      Related to a separate question we posted about the OpenLDAP/Generic LDAP directory type — this one is about ProGet's purpose-built Active Directory support instead, since we assumed it might be a separate code path.

      Our ProGet instance runs in Docker on a Linux host that is not domain-joined (a real constraint for us — we can't easily join this host to a Windows domain). We stood up a real Samba Active Directory domain controller for testing, confirmed a genuine domain user via samba-tool (sAMAccountName=aduser1, userPrincipalName=aduser1@ourtest.local), and tried to authenticate against it using both Active Directory directory types, across six configurations:

      1. V4: Active Directory/LDAP, with an explicit "Domain controller host" override (since we're not domain-joined) — bare username.
      2. Same V4 directory, after confirming it was explicitly enabled.
      3. V5: Active Directory (the newer type, with its own "Connection" tab exposing the same host-override field) — bare username.
      4. Same V5 directory — UPN-format username (aduser1@ourtest.local), in case ProGet only routes domain-qualified logins to this directory type.
      5. Discovered the bind account's Password field had silently not persisted on save — re-entered it and retested.
      6. Reconfigured our ProGet container's DNS to resolve our test domain (in case domain discovery happens before the host override is used) and retested.

      All six produced an immediate 403, and in every case, our domain controller's own connection log showed zero incoming LDAP connections — meaning ProGet never even attempted to reach it. We ruled out the obvious environmental explanations: raw TCP connectivity works, DNS resolution works from sibling containers, the directory was confirmed enabled, and ProGet's own server logs show 200 OK on both the credential save and the enable action.

      Is the "Domain controller host" override on V4/V5 meant to fully bypass the need for the ProGet host to be domain-joined, or does Active Directory support still require an actual domain join under the hood regardless of that field? If domain-joining is required, is that documented anywhere specific to Linux/Docker deployments — we want to make sure we're not missing a supported way to do this without a domain-joined Windows host.

      Thanks,
      Sai

      posted in Support
      S
      sai.pabbareddy
    • Container / Docker scanning

      One more question from the same Enterprise trial evaluation, Noncompliant container images don't seem to be blocked on pull or promotion

      Detection itself works very well: we pushed nginx:1.14.0 to a Docker feed, and ProGet correctly inventoried it — 108 packages, 66 with vulnerabilities, 521 real CVEs (via Ubuntu Security Notices).

      We then escalated one of those findings to Contain (confirmed Noncompliant in the audit view), with Feeds.AllowNoncompliantDownloads set to false at the global level — the same setting that correctly blocks NuGet/npm/PyPI packages once they're assessed Noncompliant in our testing.

      For the container image, though:

      • A fresh docker pull of the same image from the source feed succeeded anyway.
      • Promoting the same image to a second feed also succeeded.

      Layer Scanning was already enabled on the feed. We re-tested this twice more to rule out a fluke: once by reconfirming the original image's assessment was still genuinely Contain and re-pulling (same result), and once on an entirely different image (httpd:2.4.29, 95 packages, 314 vulnerabilities) end-to-end (same result).

      Is there a separate setting that governs blocking specifically for container/Docker feeds, distinct from AllowNoncompliantDownloads? Or is policy-based blocking for container images not yet supported the same way it is for package feeds on this version?
      We want to make sure we're not missing a config option before we document this as a gap in our evaluation.

      Thanks,
      Sai

      posted in Support
      S
      sai.pabbareddy
    • RE: Vulnerability scanning

      Alana,

      Thanks for the clarification on CVE-2025-26646/Category 1 — that answered our question completely.
      Following up with two more things we ran into during the same Enterprise trial evaluation, since we're trying to understand whether these are also intentional design choices or worth reporting as bugs.

      1. Severity rating disagreement vs. NVD

      System.Drawing.Common 4.5.0 → CVE-2021-24112. When first flagged in our instance, ProGet rated this "Moderate". NVD (and Trivy, which we currently run alongside our existing pipeline) rate the same CVE "Critical".

      Is ProGet's risk rating here based on an independent assessment (similar to the Category-1 reasoning you gave for CVE-2025-26646), or is it meant to track NVD's CVSS score directly and this looks like a data/sync issue? If it's an independent assessment, is there somewhere we can see the reasoning behind it, the way PGV-2535204's Category-1 write-up explained the reasoning for that one?

      1. Client-facing vulnerability warning doesn't seem to update in real time

      We changed the assessment on a specific vulnerability four times in a row (Contain → Monitor → Contain → Monitor), and re-ran a fully cache-cleared dotnet restore after each change. The NU1902 warning shown to the NuGet client did not change to reflect any of those updates — it stayed the same throughout.

      Separately, we confirmed that the actual blocking behavior (packages failing to download once assessed Noncompliant, withAllowNoncompliantDownloads=false) does react immediately/in real time to the same assessment changes. So the disconnect seems specific to the advisory/warning channel, not the enforcement channel.

      Is the client-facing NU1902 advisory feed generated on a schedule (e.g., alongside the daily vulnerability database sync) rather than reflecting live assessment state? If so, is there a way to force/trigger a regeneration, or a recommended interval to expect it to catch up on its own?

      Context: we're doing a structured Enterprise trial evaluation and documenting exactly this kind of behavior so we can give an accurate internal recommendation — appreciate you taking the time on the first question, it directly changed a conclusion in our write-up.

      Thanks,
      Sai

      posted in Support
      S
      sai.pabbareddy
    • Vulnerability scanning

      I am evaluating enterprise. Real detection engine, but its database doesn't fully overlap with NVD/Trivy, and its client-facing warnings are a periodic snapshot rather than real-time (re-verified). Live blocking, separately, is real-time.
      Coverage gap
      Microsoft.Build.Tasks.Core 17.7.2 → CVE-2025-26646 (HIGH per NVD, already known from our existing Trivy-based scanning) was not present at all in ProGet's database. Re-confirmed with the same isolated, ProGet-only source.

      posted in Support
      S
      sai.pabbareddy
    • 1
    • 2
    • 2 / 2