tests/cluster: try to catch an refresh race

In theory the `snap refresh` should have been done but it wasn't: ``` Tue, 26 Nov 2024 12:21:29 GMT + lxc exec cluster-2771-2 -- snap refresh Tue, 26 Nov 2024 12:21:29 GMT All snaps up to date. Tue, 26 Nov 2024 12:21:29 GMT + lxc exec cluster-2771-2 -- snap switch lxd --channel=5.0/edge Tue, 26 Nov 2024 12:21:29 GMT error: snap "lxd" has "refresh-snap" change in progress ``` The snapd team asked to look at `snap changes` around that issue. Signed-off-by: Simon Deziel <[email protected]>
canonical · Nov 26, 2024 · f8aa537 · f8aa537
1 parent e1a0e48
commit f8aa537
Showing 1 changed file with 2 additions and 0 deletions.
diff --git a/tests/cluster b/tests/cluster
@@ -149,6 +149,8 @@ fi
 echo "==> Upgrading the cluster"
 for i in $(seq "${SIZE}"); do
     lxc exec "${PREFIX}-$i" -- snap refresh
+    # XXX: there should be no refresh ongoing but we've seen races before so let's collect evidences
+    lxc exec "${PREFIX}-$i" -- snap changes
     lxc exec "${PREFIX}-$i" -- snap switch lxd --channel="$3"
     if [ "$i" = "${SIZE}" ]; then
         lxc exec "${PREFIX}-$i" -- snap wait system seed.loaded