summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorMiloslav Trmač <mitr@redhat.com>2021-11-29 18:58:38 +0100
committerMiloslav Trmač <mitr@redhat.com>2022-01-03 12:31:47 +0100
commitf6a3eddd2cbe6434c87cf1db04aaa56aac935e5c (patch)
tree9c3aab4119e2962ea2e0f7b924b56ebb07c6c0e0
parent90e74e794cc0c2c34112877821ed9ff0e7f51c28 (diff)
downloadpodman-f6a3eddd2cbe6434c87cf1db04aaa56aac935e5c.tar.gz
podman-f6a3eddd2cbe6434c87cf1db04aaa56aac935e5c.tar.bz2
podman-f6a3eddd2cbe6434c87cf1db04aaa56aac935e5c.zip
Don't initialize the global RNG with GinkgoRandomSeed() in e2e tests
- It probably doesn't actually make a difference: in experiments, the github.com/containers/storage/pkg/stringid RNG initialization has been happening later - This makes the RNG caller-controlled (which we don't benefit from), but also the same on all nodes of multi-process Ginkgo execution. So, if it works at all, it may make collisions of random ID values more likely, and our tests are not robust against that. So don't go out of our way to make collisions more likely. Signed-off-by: Miloslav Trmač <mitr@redhat.com>
-rw-r--r--test/utils/utils.go4
1 files changed, 0 insertions, 4 deletions
diff --git a/test/utils/utils.go b/test/utils/utils.go
index a73bdb856..1f5067950 100644
--- a/test/utils/utils.go
+++ b/test/utils/utils.go
@@ -470,10 +470,6 @@ func Containerized() bool {
return strings.Contains(string(b), "docker")
}
-func init() {
- rand.Seed(GinkgoRandomSeed())
-}
-
var randomLetters = []rune("abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ")
// RandomString returns a string of given length composed of random characters