terraform/environment: increase deadline for scheduler jobs
As of recently, enqueuall and compute-requests scheduler jobs have
nominally be failing. Both are HTTP targets with default deadline of 3
minutes. In practice, both of these endpoints take more time and hence
their jobs are designated as DEADLINE_EXCEEDED. It turns out that the
jobs were succeeding regardless, which is a behaviour observed elsewhere
too.
However, we should not rely on this. Instead, we shoud increase the
deadline to a maximum for HTTP target, which is 30 minutes. See
https://cloud.google.com/scheduler/docs/reference/rest/v1/projects.locations.jobs#Job
for more information.
Change-Id: I9ed0e4fe97f35ed39a57d09ffdc57417c5b5eeab
Reviewed-on: https://go-review.googlesource.com/c/pkgsite-metrics/+/521876
Run-TryBot: Zvonimir Pavlinovic <zpavlinovic@google.com>
Reviewed-by: Maceo Thompson <maceothompson@google.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
diff --git a/terraform/environment/worker.tf b/terraform/environment/worker.tf
index 1fb63a2..8f1c263 100644
--- a/terraform/environment/worker.tf
+++ b/terraform/environment/worker.tf
@@ -264,6 +264,7 @@
time_zone = local.tz
project = var.project
+ attempt_deadline = "1800s" # 30 min max deadline for HTTP target
http_target {
http_method = "GET"
uri = "${local.worker_url}/compute-requests"
@@ -283,6 +284,7 @@
time_zone = local.tz
project = var.project
+ attempt_deadline = "1800s" # 30 min max deadline for HTTP target
http_target {
http_method = "GET"
uri = "${local.worker_url}/govulncheck/enqueueall?min=0"
@@ -301,6 +303,7 @@
time_zone = local.tz
project = var.project
+ attempt_deadline = "1800s" # 30 min max deadline for HTTP target
http_target {
http_method = "GET"
uri = "${local.worker_url}/govulncheck/enqueue?mode=compare&min=0"
