(0.3.0) Postgres plugin: canonicalize & sanitize adress

[donaldguy]

Rebase of #486 as requested.

As I said in #486, its tempting to think maybe we should just make the key and values
(user, sslmode, etc.) individual tags. What do you think?

[donaldguy]

Let the record show that we made literally identical changes to the postgresql_test.go except I added the metric a second time as a comment. (yours was committed between when I rebased and pushed)

diff --git a/plugins/postgresql/postgresql_test.go b/plugins/postgresql/postgresql_test.go
index 0f4ff55..566354e 100644
--- a/plugins/postgresql/postgresql_test.go
+++ b/plugins/postgresql/postgresql_test.go
@@ -57,7 +57,7 @@ func TestPostgresqlGeneratesMetrics(t *testing.T) {
        for _, metric := range intMetrics {
                _, ok := availableColumns[metric]
                if ok {
-                       assert.True(t, acc.HasIntField("postgresql", metric))
+                       assert.True(t, acc.HasIntField("postgresql", metric), metric)
                        metricsCounted++
                }
        }
@@ -65,7 +65,7 @@ func TestPostgresqlGeneratesMetrics(t *testing.T) {
        for _, metric := range floatMetrics {
                _, ok := availableColumns[metric]
                if ok {
-                       assert.True(t, acc.HasFloatField("postgresql", metric))
+                       assert.True(t, acc.HasFloatField("postgresql", metric), metric)
                        metricsCounted++
                }
        }

Anyway I'll drop my commit of that now.

[sparrc]

is this really a breaking change?

[donaldguy]

It's definitely not breaking in Telegraf proper; but it could certainly be breaking for tooling built around the data in influx.

but with the field aggregation its a kinda moot point vs 0.2

[sparrc]

looks good! I would say forget about verbatim address tags, users really shouldn't want to have their password within the tag value

[donaldguy]

So is 0.10.0 the new 0.3? should I retarget to master? how do you want to proceed ?

[sparrc]

yes, 0.10.0 is the new 0.3.0, but the commits are identical. You could target to master, as that has now been merged in

[sparrc]

@donaldguy any chance you could make your changes on a new branch and send a new PR?

[sparrc]

closing for lack of updates, @donaldguy I'll review and merge if you submit a new, rebased PR.
Thanks!