DRAFT – DRAFT – DRAFT – DRAFT – DRAFT

Introduction

File-Tracker is a home-grown system that watches for changes in files in the /wiki/spaces/DPPPD/pages/27197704 and reports on any such changes that it finds. It runs periodic fixity checks (checksums) on all files in the preservation storage volumes.

Governance:

This service was created in response to the loss of similar functionality inside a former repository tool.

It is worth noting that while fixity checking is considered to be a core requirement for preservation storage, the current repository platforms do not perform periodic fixity checks and thus we have to continue to develop and maintain File-Tracker.

Mission and Scope:

File-Tracker is used to watch the Preservation Storage Environment files and report on any changes it finds. The assumption is that such file-changes (checksum changes) are likely to indicate damaged files or "bit-rot" – by catching those errors sooner, we have a better chance to create another copy from the still-good backup copies in the PSE.

Stakeholders:

John Pormann, Core Services
Giao Luong Baker, DPC
Matthew Farrell, RL/University Archives
DP3 Committee, including DPWG

Owners:

Functional Owner: DPWG
Technical Owner: David Chandek-Stark wrote the initial File-Tracker application and is the most knowledgeable in its operation
OIT owns and operates the primary and secondary storage volumes

Users:

DPC and RL staff directly read and write files onto primary storage, and would be the first ones to notice if a file was missing or "bad" and needed to be retrieved from secondary storage
DDR, RDR, and Dspace applications read and write files onto primary storage

Maintenance and Sustainability:

file-tracker-01.lib.duke.edu

Description	This VM runs a custom Ruby-on-Rails application that presents a straightforward web-UI to users. In the background, it is running checksum jobs on all files – it walks each file system (a lengthy process). It also mounts a shared (backend) log-storage area that the /wiki/spaces/LIB/pages/34907272 write to – this data is not integrated with other File-Tracker data, but at least users could track down where/when a file-error occurred.
Support	IT-Core Services developed the app and supports/maintains it.	https://gitlab.oit.duke.edu/dul-its/file-tracker
Community	There is no "community" for this particular service.
Strategy	The technology team is responsible for monitoring of security patches, etc.
Sunsetting	Given the time horizon for preservation of assets deposited with the libraries, any sunsetting effort must occur concurrently with a plan for data migration to another preservation and publication system. Future repository platforms could implement fixity checking internally, reducing the need for File-Tracker, but we would need all repository systems to handle their own fixity checking before sunsetting File-Tracker – including the Dark Archives (which are currently just a file system).
Risks	The current File-Tracker implementation relies on periodic "walks" of entire file-tree for each of the storage volumes. As the repository size grows, the time to perform those walks increases and could mean that we cannot walk the file system in a reasonable time-frame. Similarly, large files take considerable time to checksum, leading to apparent "hangs" in the system.

Issue Submission and Escalation:

Any user may submit an issue via the IT Support @ Duke University Libraries web form. In response to the prompt What can we help you with?, specify Servers & Systems Operations.
Members of the stakeholder group may reach the technical team on Slack at #devops
The technical team will use Gitlab issue boards to track deeper problems or issues with the system (https://gitlab.oit.duke.edu/devops/devops-planning/-/boards/734)

Review:

Annual:

The stakeholders team will meet annually with the technology team to:

Review the existing service description, stakeholder list, support structure,
Close out any issues that may no longer be relevant,
Estimate any project efforts or needs for the coming year,
Review and adjust the list of users with access to the system, including both privileged and non-privileged access control groups, in Grouper/Group Manager as well as programmatic access.

This activity falls in the November-December time frame. Any actionable work will be documented and reviewed by DST-LT and DP3.

Servers/Service Details:

file-tracker-01.oit.duke.edu	main VM running the file-tracker code
file-tracker.oit.duke.edu	DNS entry