If we mount cephfs by the recover_session option,
__do_request can return directly until the client automatically reconnects.
Signed-off-by: Yanhu Cao <[email protected]>
---
fs/ceph/mds_client.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 486f91f9685b..e6cda256b136 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -2708,6 +2708,12 @@ static void __do_request(struct ceph_mds_client *mdsc,
put_request_session(req);
+ if (mdsc->fsc->blacklisted &&
+ ceph_test_mount_opt(mdsc->fsc, CLEANRECOVER)) {
+ err = -EACCES;
+ goto finish;
+ }
+
mds = __choose_mds(mdsc, req, &random);
if (mds < 0 ||
ceph_mdsmap_get_state(mdsc->mdsmap, mds) < CEPH_MDS_STATE_ACTIVE) {
--
2.24.2 (Apple Git-127)
On Fri, 2020-04-17 at 17:36 +0800, Yanhu Cao wrote:
> If we mount cephfs by the recover_session option,
> __do_request can return directly until the client automatically reconnects.
>
> Signed-off-by: Yanhu Cao <[email protected]>
> ---
> fs/ceph/mds_client.c | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
> index 486f91f9685b..e6cda256b136 100644
> --- a/fs/ceph/mds_client.c
> +++ b/fs/ceph/mds_client.c
> @@ -2708,6 +2708,12 @@ static void __do_request(struct ceph_mds_client *mdsc,
>
> put_request_session(req);
>
> + if (mdsc->fsc->blacklisted &&
> + ceph_test_mount_opt(mdsc->fsc, CLEANRECOVER)) {
> + err = -EACCES;
> + goto finish;
> + }
> +
> mds = __choose_mds(mdsc, req, &random);
> if (mds < 0 ||
> ceph_mdsmap_get_state(mdsc->mdsmap, mds) < CEPH_MDS_STATE_ACTIVE) {
We do have an EBLACKLISTED error defined, which is the same as
ESHUTDOWN. The read and write code can return that when the client is
blacklisted. Same for cap handling (in __ceph_pool_perm_get. Should this
return -EBLACKLISTED instead?
--
Jeff Layton <[email protected]>
Yeah, patch the next commit.
On Fri, Apr 17, 2020 at 6:41 PM Jeff Layton <[email protected]> wrote:
>
> On Fri, 2020-04-17 at 17:36 +0800, Yanhu Cao wrote:
> > If we mount cephfs by the recover_session option,
> > __do_request can return directly until the client automatically reconnects.
> >
> > Signed-off-by: Yanhu Cao <[email protected]>
> > ---
> > fs/ceph/mds_client.c | 6 ++++++
> > 1 file changed, 6 insertions(+)
> >
> > diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
> > index 486f91f9685b..e6cda256b136 100644
> > --- a/fs/ceph/mds_client.c
> > +++ b/fs/ceph/mds_client.c
> > @@ -2708,6 +2708,12 @@ static void __do_request(struct ceph_mds_client *mdsc,
> >
> > put_request_session(req);
> >
> > + if (mdsc->fsc->blacklisted &&
> > + ceph_test_mount_opt(mdsc->fsc, CLEANRECOVER)) {
> > + err = -EACCES;
> > + goto finish;
> > + }
> > +
> > mds = __choose_mds(mdsc, req, &random);
> > if (mds < 0 ||
> > ceph_mdsmap_get_state(mdsc->mdsmap, mds) < CEPH_MDS_STATE_ACTIVE) {
>
> We do have an EBLACKLISTED error defined, which is the same as
> ESHUTDOWN. The read and write code can return that when the client is
> blacklisted. Same for cap handling (in __ceph_pool_perm_get. Should this
> return -EBLACKLISTED instead?
>
> --
> Jeff Layton <[email protected]>
>