GitHub Backup

[plehegar]

We do backup repos and their data. Not sure how well it is documented however. Needs some facts finding.

[tripu]

I can point you in the way of our current approach to GH backups.

(BTW, we have this public project, gh-backup; almost two years without maintenance. Either we take it up, or we discontinue it — and make that clear on the project page.)

[dontcallmedom]

Having looked at the backups, they are indeed incomplete when it comes to non-git data:

• we only have the latest 30 issues per repo (corresponding to the default pagination of the github API)
• we only capture data about issues, nothing about the comments made on the issues (where most of the value actually resides)

It would probably be useful to take a thorough review of the tool we use for the backup, and in particular compare it with the many data that github exposes via its API: https://developer.github.com/v3/

[plehegar]

I wonder, don't we have the issues/comments through pheme?

[dontcallmedom]

we have a subset of the data on issue and comments via pheme; but pheme is not a backup system and is not built for long-term reliability.

[xfq]

The User Migration API announced recently seems helpful.

[r12a]

Could the systems team please clarify for the staff the current situation wrt github backups, and say something about plans for ongoing improvements. I thought i heard last week that we have backups of most of the data, but not in an accessible form. It sounds like we should be working on that as a pretty high priority, given that much of our institutional knowledge is affected.

[vivienlacourba]

Hi @r12a, W3C Systeam already has backups of our GitHub data in place. We will communicate on this soon.