PromQL tests fail to run on a system with 4gb /tmp space

[grobie]

What did you do?

go test github.com/prometheus/prometheus/promql

What did you expect to see?

Tests to succeed.

What did you see instead? Under which circumstances?

...
--- FAIL: TestEvaluations (7.40s)
        promql_test.go:33: error running test testdata/functions.test: log series: no space left on device
        promql_test.go:33: error running test testdata/histograms.test: log series: no space left on device
        promql_test.go:33: error running test testdata/legacy.test: log series: no space left on device
        promql_test.go:33: error running test testdata/operators.test: log series: no space left on device
        promql_test.go:33: error running test testdata/selectors.test: log series: no space left on device
        promql_test.go:33: error running test testdata/staleness.test: log series: no space left on device
FAIL
FAIL    github.com/prometheus/prometheus/promql 7.861s

Environment

• System information:

  Linux 4.13.6-1-ARCH x86_64

tmpfs is 4GB: tmpfs on /tmp type tmpfs (rw,nosuid,nodev)

• Prometheus version:

  master + updated tsdb (commit: 721050c)

[gouthamve]

This has happened with me too but I think I had /tmp with 2G (away from my machine). I think the issue is not that a test can produce 4G but the artifacts from the older tests are not being cleaned. Needs to be investigated, but I think a defer os.RemoveAll() can fix this.

[grobie]

I actually deleted everything in /tmp and started with 99% free space and still couldn't run the tests.

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.