Return list of ongoing task in /health endpoint

[mangalaman93]

We return the list of ongoing tasks out of snapshot, indexing or/and rollup in the health endpoint. This endpoint can now be used to check whether indexing is still going on after an Alter operation completes.

TODO

• add docs

This change is 

Docs Preview:

[pawanrawal]

Could we add a test case for this?

Also please verify that this info should also be returned from the GraphQL endpoint. You'd just need to add the appropriate type for the new field inside GraphQL schema at graphql/admin/admin.go. Happy to help with it.

Reviewed 6 of 6 files at r1.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @mangalaman93, @manishrjain, and @parasssh)

---

worker/draft.go, line 154 at r1 (raw file):

			return "snapshot"
		case opIndexing:
			return "indexing"

Would also be nice to get info about the predicate for which indexing is going on. Can indexing for multiple predicates go on at the same time?

[manishrjain]

Let's get info from all the other servers via Heartbeat.

Reviewed 6 of 6 files at r1.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @mangalaman93 and @parasssh)

[parasssh]

Reviewable status: all files reviewed, 2 unresolved discussions (waiting on @mangalaman93)

---

worker/draft.go, line 166 at r1 (raw file):

		tasks = append(tasks, toStr(id))
	}
	return tasks

Since we are only trying to indicate that some "indexing, rollup and/or snapshot" is ongoing, it may be good to remove dupes before returning.

Alternatively, we add more info per task such as predicate being indexed, snapshot ts etc.

[mangalaman93]

There is a cyclic dependency in providing tasks information in heartbeats.
worker -> conn -> worker (for tasks information)

Reviewable status: 0 of 9 files reviewed, 2 unresolved discussions (waiting on @manishrjain, @parasssh, and @pawanrawal)

---

worker/draft.go, line 154 at r1 (raw file):

Previously, pawanrawal (Pawan Rawal) wrote…

Would also be nice to get info about the predicate for which indexing is going on. Can indexing for multiple predicates go on at the same time?

Done.

---

worker/draft.go, line 166 at r1 (raw file):

Previously, parasssh wrote…

Since we are only trying to indicate that some "indexing, rollup and/or snapshot" is ongoing, it may be good to remove dupes before returning.

Alternatively, we add more info per task such as predicate being indexed, snapshot ts etc.

A map cannot have duplicates to begin with. Added more info

[pawanrawal]

is this something that we could have an automated test case for?

Reviewed 9 of 9 files at r2.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @parasssh)

[manishrjain]

Reviewed 9 of 9 files at r2.
Reviewable status: all files reviewed, 2 unresolved discussions (waiting on @mangalaman93 and @parasssh)

---

edgraph/server.go, line 725 at r2 (raw file):

		Uptime:   int64(time.Since(x.WorkerConfig.StartTime) / time.Second),
		LastEcho: time.Now().Unix(),
		Ongoing:  worker.GetOngoingTasks(),

Add a TODO that we need to find a way to get the health from others.

[mangalaman93]

@pawanrawal the test could become flaky because indexing won't take too long if the data size is small.

Reviewable status: all files reviewed, 2 unresolved discussions (waiting on @manishrjain and @parasssh)

---

edgraph/server.go, line 725 at r2 (raw file):

Previously, manishrjain (Manish R Jain) wrote…

Add a TODO that we need to find a way to get the health from others.

Done.