Open mbarnes opened 7 years ago
You should do
systemctl stop docker
atomic storage reset
atomic storage modify --storage=overlay2
systemctl start docker
I should have emphasized this is when running atomic storage reset
again after all that.
Observed this while working on a devicemapper-to-overlay2 playbook (based on yours) and trying to reset the state back to devicemapper (see https://github.com/openshift/openshift-ansible/pull/5216).
@rhvgoyal Ideas?
What about just:
diff --git a/Atomic/storage.py b/Atomic/storage.py
index 77c1cbb..6e02256 100644
--- a/Atomic/storage.py
+++ b/Atomic/storage.py
@@ -160,6 +160,7 @@ class Storage(Atomic):
util.call(["umount", root + "/devicemapper"], stderr=DEVNULL)
util.call(["umount", root + "/overlay"], stderr=DEVNULL)
util.call(["umount", root + "/overlay2"], stderr=DEVNULL)
+ util.call(["umount", root], stderr=DEVNULL)
shutil.rmtree(root)
os.mkdir(root)
try:
And let it silently fail if there's nothing mounted there?
If that works, open a pull request with atomic.
Gah, just realized I filed this issue against the wrong project. Sorry about that. 😖
Following the RHEL documentation for setting up the
overlay2
graph driver, which recommends:atomic storage reset
subsequently fails withbecause the LV is still mounted there.