db: improve compaction test timeout observability

[RaduBerinde]

• set up a 1 minute timeout for any datadriven compactions;
• print db logs on test failures;
• upgrade datadriven to include a recent change to periodically log
  when a single command takes a long time.

Informs #5277

[cockroach-teamcity]

This change is 

[RaduBerinde]

(pending merge of cockroachdb/datadriven#57)

[xxmplus]

Reviewable status: 0 of 12 files reviewed, 1 unresolved discussion (waiting on @jbowens)

---

compaction_test.go line 986 at r1 (raw file):

		defer func() {
			if t.Failed() {
				t.Logf("db log:\n%s\n", dbLog.String())

unrelated newbie question: do we print severity in logs in Pebble? or do we leave it to crdb to do that? I looked it up and it seems golang support structured error message via https://pkg.go.dev/log/slog.

[RaduBerinde]

TFTR!

Reviewable status: 0 of 12 files reviewed, 1 unresolved discussion (waiting on @jbowens and @xxmplus)

---

compaction_test.go line 986 at r1 (raw file):

Previously, xxmplus (Andy Xu) wrote…

unrelated newbie question: do we print severity in logs in Pebble? or do we leave it to crdb to do that? I looked it up and it seems golang support structured error message via https://pkg.go.dev/log/slog.

The default implementation (below) doesn't, and neither does this test-only InMemLogger. CRDB implements its own logger which does take into account the type of call (Info/Error/Fatal).

pebble/internal/base/logger.go

Line 32 in ac13f40

func (defaultLogger) Infof(format string, args ...interface{}) {