Justify the handlers tests' start limit with a measurement (closes #225)

The internal/handlers tests were load-fragile because every test app hashed the admin password at 64 MB; that went with the cheaper test hashing already on next, and measuring under the host's real load found nothing left to fix in how the tests run. The comment on newTestApp now says its start limit, fx's default, is there to catch a start that hangs, and that the slowest measured start is far inside it. The header of script/test gives current figures in place of ones from before that change. Neither limit changes, and no test changes. Model: opus-5-5
2026-10-02 14:08:54 +02:00
parent 9ade217222
commit 8cf5acaf1d
2 changed files with 14 additions and 3 deletions
@@ -2,9 +2,10 @@
 # script/test: run the test suite.
 #
 # -timeout is applied by `go test` per package, not to the run as a whole, so
-# it only has to clear the slowest single package. That is internal/handlers,
-# measured in a cache-defeated builder stage on the 48-core shared build host
-# (2026-08-18); load- and host-dependent, not invariants:
+# it only has to clear the slowest single package. When this budget was set
+# that was internal/handlers, measured in a cache-defeated builder stage on the
+# 48-core shared build host (2026-08-18); load- and host-dependent, not
+# invariants:
 #
 #   16.9s                  host load 5-20, GOMAXPROCS 48
 #   45.9s / 47.3s / 49.0s  three runs at deliberate host load 31-73
@@ -23,6 +24,11 @@
 # a condition CI runs under. If a CPU-limited runner ever puts a real run near
 # 67s, that is the datum to revisit the org figure with.
 #
+# Those figures predate tests hashing the admin password at 1 MB instead of
+# 64 MB (https://git.eeqj.de/sneak/webhooker/pulls/404). After that change, in
+# a cache-defeated build at host load 44-109 (2026-10-02), internal/handlers
+# took 8.5s and the slowest package was internal/database at 15.8s.
+#
 # -p 4 -parallel 8 keep the run under 2 GB of memory: at most four test
 # binaries build or run at once, each with at most eight parallel tests. Under
 # -race every test binary and every link costs a few hundred MB, so the