How to hard-update a Graphryder dashboard

There is a trigger to update the Graphryder API. By “update” we mean here re-import the data from Discourse, and place it in its own Neo4J DB.

It works like this:

1. Make sure the relevant content is tagged for harvesting

Graphryder collects data on the basis of a (Discourse) tag identifying the project. By convention, these tags always start in ethno-. For NGI Forward, for example, it is ethno-ngi-forward.

To make sure you have the right collection of topics, you need to tag each topic with the project’s tag. Most projects in Edgeryders have their own cats: for NGI Forward, the cat is #ioh. Occasionally community managers and ethnographer assign an ethno- tag to some content that lives elsewhere (for example in past projects), but normally every top in the project cat is part of that project’s ethnographic corpus. This means, every top in the project cat needs to be tagged with the project’s tag as well.

To do this:

  • Go through all public sub-categories plus the /none sub cat (example
  • For each of these sub-cats, scroll down until the end of its topic lists
  • Next, go back to the top and click on the Topics button to the left:
  • Click on the Select all button. A wrench icon appears on the right. Click it.
  • Click Append tag, then select or type the ethno- tag as appropriate.
    Select all topics and append the tag ethno-ngi-forward

2. Re-import your data into Graphryder’s database.

@hugi says:

Just visit:

Never do this just before a demo though as it takes about 30 minutes to rebuild.

3. Regenerate the graph

After it’s done (you’ll see the numbers tick up as you refresh the dashboard), go to the dashboard, open the right hand menu and click all the graph regeneration toggles.

The hard update does not work reliably. Right now I am seeing a 502 error. Full traceback in the file attached.

error_502.txt (5.1 KB)

It looks like Neo4j is choking. Maybe it’s running out of memory, I know that neo is pretty memory intense. @matthias?

Other times I get this:

… however, if I relaod the dashboard, I see the user and post counts keep going up… fingers crossed.

There’s 8 GiB of free main memory around now. It did run with 1-2 GiB free memory in the past, so I doubt we got a memory issue. But you can check of course, as you have the SSH login now.

1 Like