tasks/ceph_fuse.py: virtual machines need flexible mount timeout

1) add a wait time before the mount attempt to let the cluster get set up.
By default this should be skipped, but for VMs and known-slow systems we
can give them 60 seconds.
2) Make the timeout configurable, with a 30-second default, but override it
for VM tests.

http://tracker.ceph.com/issues/12320 Fixes: #12320

Signed-off-by: Loic Dachary <loic@dachary.org>

diff --git a/machine_types/vps.yaml b/machine_types/vps.yaml
index 0ef045b4ca7886fe0549d83d8523fb7ffb0c276e..bffa0985da8d276a0500e15717c5ae370bce7893 100644 (file)
--- a/machine_types/vps.yaml
+++ b/machine_types/vps.yaml
@@ -10,3 +10,7 @@ overrides:
     default_idle_timeout: 1200
   s3tests:
     idle_timeout: 1200
+  ceph-fuse:
+    client.0:
+       mount_wait: 60
+       mount_timeout: 120

diff --git a/tasks/ceph_fuse.py b/tasks/ceph_fuse.py
index 60988dd1fdc6beb77274359dd4e45271c2ecc632..9c308ae36cababd45be6d8e4fd63d26dc484f620 100644 (file)
--- a/tasks/ceph_fuse.py
+++ b/tasks/ceph_fuse.py
@@ -79,6 +79,16 @@ def task(ctx, config):
                     mounted: false
             - ... do something that requires the FS unmounted ...
 
+    Example that adds more generous wait time for mount (for virtual machines):
+
+        tasks:
+        - ceph:
+        - ceph-fuse:
+            client.0:
+              mount_wait: 60 # default is 0, do not wait before checking /sys/
+              mount_timeout: 120 # default is 30, give up if /sys/ is not populated
+        - interactive:
+
     :param ctx: Context
     :param config: Configuration
     """

diff --git a/tasks/cephfs/fuse_mount.py b/tasks/cephfs/fuse_mount.py
index 5517f5f8474cd0dca3bf8546a6c3c0a8eca6faa2..6332852b8232afb63e92654e9a601d40003a55d5 100644 (file)
--- a/tasks/cephfs/fuse_mount.py
+++ b/tasks/cephfs/fuse_mount.py
@@ -100,11 +100,16 @@ class FuseMount(CephFSMount):
         self.fuse_daemon = proc
 
         # Wait for the connection reference to appear in /sys
+        mount_wait = self.client_config.get('mount_wait', 0)
+        if mount_wait > 0:
+            log.info("Fuse mount waits {0} seconds before checking /sys/".format(mount_wait))
+            time.sleep(mount_wait)            
+        timeout = self.client_config.get('mount_timeout', '30')
         waited = 0
         while list_connections() == pre_mount_conns:
             time.sleep(1)
             waited += 1
-            if waited > 30:
+            if waited > timeout:
                 raise RuntimeError("Fuse mount failed to populate /sys/ after {0} seconds".format(
                     waited
                 ))