2023-07-20 Meeting notes

Owned by Craig McNally
Last updated: Jul 20, 2023
6 min read

Date

Attendees

NamePresent

Craig McNally 

Y
Julian Ladisch 
Y

Axel Dörrer 

Y
Ryan Berger 

Chris Rutledge 

Y

Jakub Skoczen 


John Coburn 
Y

Skott Klebe 




Discussion items

TimeItemWhoNotes
0 minOWASP/SNYKTeam
  • Epic: FOLIO-3582 - Getting issue details... STATUS
  • Feature: FOLIO-3583 - Getting issue details... STATUS
  • User Story: FOLIO-3584 - Getting issue details... STATUS
  • User Story:  FOLIO-3709 - Getting issue details... STATUS
  • Sonar ( https://sonarcloud.io/organizations/folio-org/quality_profiles):
    • Java: All 38 security hotspots rules and all 53 vulnerabilities rules are enabled (2 deprecated vulnerabilities rules are disabled).
    • JavaScript: 53 security hotspots rules and all 27 vulnerabilities rules are enabled.
      • John Coburn will check whether the remaining 2 security hostspots rules should be enabled.
  • Snyk: Skott Klebe to take a closer look and add ignore where applicable

Today:

  • What about the OWASP stuff?
1 minNCT group (Pen. testing)

Axel Dörrer 

Progress is slow... at most expect monthly updates.

  • Pre work has been completed (Python)
  • Next step is to parse the RAML for all endpoints/module and run the tests
1 minSTCOR-395 "refactor login form to avoid using any form framework whatsoever"

All

 Background
  • Create a spike to investigate what's involved in divorcing the login page from the NPM ecosystem.  Will reach out to John and Ryan as needed. 
  • Reopen STCOR-395 and block on the spike. – Done.
  • Where does this stand?  Get an update from Ryan Berger / John Coburn
  • Waiting for the spike ( STCOR-651 - Getting issue details... STATUS ) to be completed. – currently in the Open state.
  • John to reach out to Skott to discuss what the level of risk associated with this.  – Still needs to happen.
  • John Coburn pulled together two PoCs.  See comments in STCOR-651 - Getting issue details... STATUS
  • How do we want to move forward?
    • Solutions need to be reviewed and discussed.
    • Sounds like the iframe approach is a non-starter... actually a step in the wrong direction security-wise
  • John Coburn (and others) to read up on browser CSPs
  • John Coburn has made some progress on investing CSPs
    • Will share some draft guidance we may want to include into the installation documentation (via slack)
    • SG will review and provide feedback.  Skott Klebe please take a look too
  • Next up:  John to work on some spike work - focused on introducing CSPs on the folio-snapshot site
    • can serve as a reference impl of the guidance we'll be adding to the install docs
  • Not much progress since last week, but hopefully get some movement on this soon.
  • FOLIO-3691 - Getting issue details... STATUS Spike
  • John Coburn trying to get this lined up for next sprint
  • John Coburn to discuss with another developer (Maccabee) who is familiar with CSPs.  
  • John had the chance to look on it but only via metatag need to try out or examine http header configuration at the server side
  • Some things happen in stripes modularity too which might have impact too
  • John Coburn is making progress - has done some testing locally, but wants to exercise it in a more realistic env.
  •  John Coburn provided update on CSP effort.
    • Experimenting with express to provide CSP (for local dev purposes).  Production systems wouldn't use this, it would be formalized by the hosting provider.
    • This allows for local testing and experiments. 

Today:

  • Nothing new to report this week.
1 minDisable tenant checking to support multi tenant requests (MODAT-143).

Julian Ladisch / All

A few wiki pages have been shared on this... See Enhanced Consortia Support(ECS)

Julian Ladisch will discuss his concerns with Olamide, etc. and we can discuss here again if/when needed.

  • There's a need to allow for the user to easily switch between tenant contexts
  • Maybe making this opt-in is a reasonable approach
    • the restriction would remain unchanged by default, but you could relax this constraint by making an explicit configuration change
  • Julian Ladisch met with Thunderjet/Olamide
  • mod-consortia/ui-consortia have been submitted for TC review
    • This will probably be raised as part of those reviews
  • Added a comment to the TCR-26 - Getting issue details... STATUS :
    Security team disagree with breaking the tenant seperation on okapi token level. We would like to encourage an alternative solution on saml or openid techniques which would be less invasive to the current approach.
  • TC would like to split off the security/token concerns from the module acceptance, will be discussed next Wednesday TC's meeting
  • Craig McNally  to raise this with the TC again and get the ball rolling on discussing the larger topic (e.g. via a subgroup?  RFC?  something else?)
    • Will involve Axel and Julian to a subgroup
  • The TC approved mod-consortia, but also wanted to discuss the concerns raised about tenant isolation and relaxing tenant checks in certain circumstances.  The TC plans to discuss on Monday 6/5 11:00 AM ET. → Julian and Axel to join the meeting
  • TC discussed on 2023-06-26 - Consortia Tenant Checks
  • Raised this at the TC and the sentiment is that this group should work to resolve concerns with Olamide on its own.  Need to discuss next steps... 
    • Try to find a time for Olamide to discuss with us? 
    • Try to handle this "out of band" via slack, google doc, etc.?
    • Ask Olamide to join one of our Thursday meetings?
    • Something else?

Today:

  • Not a ton of feedback on this today - Craig McNally  will reach out to Olamide to see how he'd like to proceed
5 minFOLIO-3535 Upgrade bitnami/elasticsearch:7.10.2 in reference and vagrant development  boxes (folio-ansible)All
  • Jakub Skoczen to bring this to the devops team asking to  bump elasticsearch to a major version in this case
  • but there is the concern who should responsible to keep these environments up to date and maintain them in general
  • the devops team is greatly lacking on resources and can't take the task permanently
  • need to have a discussion in a wider group (TC?)
  • Craig McNally to touch base with Jakub Skoczen about this in slack.  Get it on the TC agenda if needed.
    • My guess is that if DevOps can't do this, it will likely fall on the Kitfox team.  It should be discussed with Oleksii P. and Mark V. 
    • I don't think the TC will be helpful here since they don't direct development resources/teams/etc. 

Today:

  • Craig McNally provided an update - last I heard Ingolf was going to bring this up at the SysOps SIG.  Ingolf may be on vacation.
5-10 min

OKAPI-853 - Getting issue details... STATUS

Craig/Jakub

The wording should be adjusted, it's a little misleading

Also need to determine if this is a must have for the Refresh token work.

See discussion in slack channel for additional details.

TODO:

  • Craig McNally to create a JIRA for using SameSite: Lax (or possibly Strict) instead of "None".  See refresh token PR in mod-login.
  • Craig McNally to fix up OKAPI-853 with clarifications

Ideas:

  • Lock down by default (to the hostname which OKAPI is using), but allow additional origins to be allowed via configuration
  • UI developers often need the ability to point a locally hosted UI to a backend hosted elsewhere

Today:

  • Deferred to discuss other important topics.
*Critical Security VulnerabilitiesTeam
  • Fixes have been implemented and releases have been made for Orihid and Nolana.
  • Notifications have been sent to SysOps mailing list and #sys-ops slack channel
  • Is there any way to tell if there was unauthorized access?
    • OKAPI logs the user id for all proxied calls.  It may be possible to look for unusual activity associated with these system users
    • Load balancer or reverse proxy logs may provide some insight. 
      • Especially if the X-Forwarded-For header is logged... 
      • Other information may not be logged though, e.g. the username/id.
    • Craig McNally will respond to Jason in #sys-ops
  • What does the timing look like for full disclosure?
    • Probably at least a day or two.
    • At least give them the weekend so there's a good window for downtime, etc.
    • Craig McNally will ask the sys-ops community.
  • Still need to draft a statement for full disclosure
  • Where does the full disclosure statement get sent?  #sys-ops, #folio-general?  multiple channels?  an email distro?
    • Try to consolidate the conversation in a single place (#folio-general)
  • We will need to adjust the Security Level on the related JIRAs once we fully disclose this
  • GitHub advisories will need to be published when we fully disclose this

*

Review the Kanban board. Team
  • Reviewed tickets which haven't moved recently.  We made it up through EDGERTAC-72, then ran out of time.

Today:

  • MODHAADM-27 - Getting issue details... STATUS
    • PO Charlotte Whitt  requested: "While we need to be careful about spending our available developer time wisely, it would be helpful if you could indicate whether this work is a must fix now, or something we can fix when/if we decide to build the FOLIO Harvester module."
    • https://github.com/indexdata/localindices/pull/90 "MODHAADM-27: Migrate log4j from 1 to 2#90". This pull request uses the Log4j 1.x bridge (log4j-1.2-api), no code change is needed. This fixes all known log4j security issues, including https://nvd.nist.gov/vuln/detail/CVE-2019-17571 .
    • What advice does the security team give? Fix now or postpone to 2024?
Topic Backlog

Retiring issues which have been open for a long time w/o progressAll

Discussed gathering a report for the TC to review/approve.  Need to work out details/logistics.

Query so far:

  • labels = Security AND created < '-52w' AND status != closed AND status != completed AND status != Cancelled

Bot Detection/ControlAll
  • Not a huge problem due to needing to authenticate first, and the user has the required permissions to get the information sought after.
  • If using AWS, WAF bot control may provide some protection w/ little effort - Skott Klebe to investigate.
  • Craig McNally to check in with Skott on this

Time slotAllDo we need a better time slot for the security team meeting to allow more members to join?

Logging & Personal DataCraig/Team

A developer recently reached to me asking if the security team or TC has guidance or rules in place for logging of personal data.  Some guidelines are documented on the wiki, but I'm wondering if it's worth making some clarifications and creating a draft decision record for the TC to formally endorse

Is this even in our purview?  Should we seek input from the Privacy SIG?  Should I raise this with the TC first?

Next steps:

  • For now, put this on hold.  See how the Draft DR approach works for the periodic dependency updates (see above).  If that goes smoothly, we'll take this on next.  Otherwise we'll consider other approaches.

Cyber Resilience ActTeam

Action items

  •