extend tempest-integrated-compute and tempest-full-py3

This change adds 30mins to the timeout to account
for timeouts on slow nodes and gather more data.

once stabilised we should consider
enabling zswap, or adding more swap to account
for OOM issues thereby allowing a higher concurrency
to reduce overall execution time.

Change-Id: I9529f554acb5b67401ed1cb8700a916c7131ec02
diff --git a/zuul.d/integrated-gate.yaml b/zuul.d/integrated-gate.yaml
index 2fd6e36..1343a7c 100644
--- a/zuul.d/integrated-gate.yaml
+++ b/zuul.d/integrated-gate.yaml
@@ -65,6 +65,10 @@
     branches:
       regex: ^.*/(victoria|wallaby)$
       negate: true
+    # NOTE(sean-k-mooney): this job and its descendants frequently times out
+    # run on rax-* providers with a timeout of 2 hours. temporary increase
+    # the timeout to 2.5 hours.
+    timeout: 9000
     description: |
       Base integration test with Neutron networking, horizon, swift enable,
       and py3.
@@ -78,6 +82,8 @@
       # end up 6 in upstream CI. Higher concurrency means high parallel
       # requests to services and can cause more oom issues. To avoid the
       # oom issue, setting the concurrency to 4 in this job.
+      # NOTE(sean-k-mooney): now that we use zswap we should be able to
+      # increase the concurrency to 6.
       tempest_concurrency: 4
       tox_envlist: integrated-full
       devstack_localrc:
@@ -133,11 +139,18 @@
       This job runs integration tests for compute. This is
       subset of 'tempest-full-py3' job and run Nova, Neutron, Cinder (except backup tests)
       and Glance related tests. This is meant to be run on Nova gate only.
+    # NOTE(sean-k-mooney): this job and its descendants frequently times out
+    # when run on rax-* providers, recent optimizations have reduced the
+    # runtime of the job but it still times out. temporary increase the
+    # timeout to 2.5 hours.
+    timeout: 9000
     vars:
       # NOTE(gmann): Default concurrency is higher (number of cpu -2) which
       # end up 6 in upstream CI. Higher concurrency means high parallel
       # requests to services and can cause more oom issues. To avoid the
       # oom issue, setting the concurrency to 4 in this job.
+      # NOTE(sean-k-mooney): now that we use zswap we should be able to
+      # increase the concurrency to 6.
       tempest_concurrency: 4
       tox_envlist: integrated-compute
       tempest_exclude_regex: ""