features/quota: Read the limit from the xattrs

Old approach:
This is based on the path. So, the config file stored as limit-set key value
pair in the volfile was the source of truth for any directory to be enforced
the quota limit. So, we were storing it in a list and everypath is compared
with the list. This hinders scaling with the order of kilos of quota
directories. This approach is different from the XFS where the below method of
inode based limit is employed this adds the admins to learn new stuff.

New approach:
This is based on the inode. After the path is set with the quota limit. The
limit sticks with the inode. So, we need not to store the paths in the xlator
context/options, instead get the xattrs in the lookup. Set the hard and soft
limits of context from there.

The additional change in the glusterd to set the xattrs when limit-usage
command issued will be submitted as a separate patch.

This patch also does:
* Move the logging from the quotad to quota-enforcer.
* Save the soft limit in the context.
* Some clean up in quota-enforcer

Change-Id: I2840f9377a4a1f76630741550873fe74758fa100
Signed-off-by: Varun Shastry <vshastry@redhat.com>

1 files changed, 3 insertions, 3 deletions

diff --git a/xlators/features/quota/src/quotad.c b/xlators/features/quota/src/quotad.c
index a5e51add..b1194870 100644
--- a/xlators/features/quota/src/quotad.c
+++ b/xlators/features/quota/src/quotad.c
@@ -560,10 +560,10 @@ qd_iterator (qd_vols_conf_t  *conf, xlator_t *subvol)
 {
         limits_t        *entry          = NULL;
         limits_t        *next           = NULL;
-        int32_t          ret;
+        int32_t         ret             = -1;
         dict_t          *dict_req       = NULL;
-	struct timeval	now;
-	struct timeval	next_expire = {0,};
+	struct timeval	now             = {0,};
+	struct timeval	next_expire     = {0,};
 
         GF_VALIDATE_OR_GOTO ("qd-iterator", conf, out);
         GF_VALIDATE_OR_GOTO ("qd-iterator", subvol, out);