Re: [libvirt] [PATCH] storage: Attempt to refresh volume after successful wipe volume
On Wed, Dec 02, 2015 at 03:22:38PM -0500, John Ferlan wrote:
https://bugzilla.redhat.com/show_bug.cgi?id=1270709
When a volume wipe is successful, a volume refresh should be done afterwards
to update any volume data that may be used in future volume commands, such as
volume resize. For a raw file volume, a wipe would truncate the file and
a followup volume resize the capacity may fail because the volume target
allocation isn't updated to reflect the wipe activity.
I would expect that after wiping a 200 MB volume with zeros, it would
contain 200 MB of zeros and it would not be shrinkable to 50 MB, not
even after a volume refresh.
While it seems ftruncate to 0 bytes and back satisfies the documentation
of the virStorageVolWipe API:
Ensure data previously on a volume is not accessible to future reads
the ALG_ZERO description says:
VIR_STORAGE_VOL_WIPE_ALG_ZERO = 0 1-pass, all zeroes
Do we need to update it to reflect that there might not be any pass over
the old data (which might not happen for non-sparse files either,
if the filesystem does not overwrite the same sectors)?
Signed-off-by: John Ferlan <jferlan(a)redhat.com>
---
src/storage/storage_driver.c | 14 +++++++++++++-
1 file changed, 13 insertions(+), 1 deletion(-)
diff --git a/src/storage/storage_driver.c b/src/storage/storage_driver.c
index bbf21f6..2e59e39 100644
--- a/src/storage/storage_driver.c
+++ b/src/storage/storage_driver.c
@@ -2436,7 +2436,19 @@ storageVolWipePattern(virStorageVolPtr obj,
goto cleanup;
}
- ret = backend->wipeVol(obj->conn, pool, vol, algorithm, flags);
+ if ((ret = backend->wipeVol(obj->conn, pool, vol, algorithm, flags)) < 0)
+ goto cleanup;
More readable as:
if (func() < 0)
goto cleanup;
ret = 0;
If the return value does not need to be propagated. (Which it should not
here, but it is possible in some cases. I will send a patch)
+
+ /* Best effort to refresh the volume data. If unsuccessful, we've already
+ * wiped the data so there's no going back on that. Best we can do is
+ * provide some details over what happened and move on
+ */
The wipe did happen, but if refreshVol fails, there is something
seriously wrong with the volume. I think returning -1 and reporting an
error is reasonable here.
Jan
+ if (backend->refreshVol &&
+ backend->refreshVol(obj->conn, pool, vol) < 0) {
+ VIR_WARN("failed to refresh volume '%s' info after volume
wipe",
+ vol->name);
+ virResetLastError();
+ }
Attachments:
- signature.asc (application/pgp-signature — 819 bytes)