On Mon, Dec 9, 2013 at 8:02 PM, Dzianis Huznou <dzianis_huznou@xxxxxxxx> wrote: > On Sat, 2013-12-07 at 21:59 +0800, Yan, Zheng wrote: >> On Sun, Dec 8, 2013 at 2:59 AM, Mikhail Campos Guadamuz >> <plageat90@xxxxxxxxx> wrote: >> > For http://tracker.ceph.com/issues/4386 >> > >> > It determines situation, when a user is trying to mount CephFS >> > with no MDS present. Return ECOMM from >> > open_root_dentry which can be analyzed then by ceph.mount >> > >> > Signed-off-by: Mikhail Campos Guadamuz <plageat90@xxxxxxxxx> >> > --- >> > fs/ceph/mdsmap.c | 19 ++++++++++++++++--- >> > fs/ceph/super.c | 10 +++++++++- >> > include/linux/ceph/mdsmap.h | 1 + >> > 3 files changed, 26 insertions(+), 4 deletions(-) >> > >> > diff --git a/fs/ceph/mdsmap.c b/fs/ceph/mdsmap.c >> > index 132b64e..3a6ba8a 100644 >> > --- a/fs/ceph/mdsmap.c >> > +++ b/fs/ceph/mdsmap.c >> > @@ -12,6 +12,20 @@ >> > >> > #include "super.h" >> > >> > +/* >> > + * count active mds's >> > + */ >> > +int ceph_mdsmap_active_mds_count(struct ceph_mdsmap *m) >> > +{ >> > + int n = 0; >> > + int i; >> > + >> > + for(i = 0; i < m->m_max_mds; ++i) >> > + if(m->m_info[i].state > 0) >> > + ++n; >> > + >> > + return n; >> > +} >> > >> > /* >> > * choose a random mds that is "up" (i.e. has a state > 0), or -1. >> > @@ -26,9 +40,8 @@ int ceph_mdsmap_get_random_mds(struct ceph_mdsmap *m) >> > return 0; >> > >> > /* count */ >> > - for (i = 0; i < m->m_max_mds; i++) >> > - if (m->m_info[i].state > 0) >> > - n++; >> > + n = ceph_mdsmap_active_mds_count(m); >> > + >> > if (n == 0) >> > return -1; >> > >> > diff --git a/fs/ceph/super.c b/fs/ceph/super.c >> > index 6627b26..4d33d68 100644 >> > --- a/fs/ceph/super.c >> > +++ b/fs/ceph/super.c >> > @@ -674,7 +674,15 @@ static struct dentry *open_root_dentry(struct ceph_fs_client *fsc, >> > struct ceph_mds_request *req = NULL; >> > int err; >> > struct dentry *root; >> > - >> > + >> > + /* check for mds*/ >> > + if( 0 == ceph_mdsmap_active_mds_count(mdsc->mdsmap) ) >> > + { >> > + pr_info("active mds not found, possible not exist\n"); >> > + root = ERR_PTR( -ECOMM ); >> > + return root; >> > + } >> >> I think we should return error immediately only if there is no mds >> (mdsmap->m_max_mds == 0). If there is inactive mds, we should stick >> to current behavior (wait until timeout). Furthermore please don't use >> uncommon error code ECOMM, use common error code such as EINVAL. >> >> Regards >> Yan, Zheng >> >> >> > + >> > /* open dir */ >> > dout("open_root_inode opening '%s'\n", path); >> > req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_GETATTR, USE_ANY_MDS); >> > diff --git a/include/linux/ceph/mdsmap.h b/include/linux/ceph/mdsmap.h >> > index 87ed09f..4d7d502 100644 >> > --- a/include/linux/ceph/mdsmap.h >> > +++ b/include/linux/ceph/mdsmap.h >> > @@ -56,6 +56,7 @@ static inline bool ceph_mdsmap_is_laggy(struct ceph_mdsmap *m, int w) >> > return false; >> > } >> > >> > +extern int ceph_mdsmap_active_mds_count(struct ceph_mdsmap *m); >> > extern int ceph_mdsmap_get_random_mds(struct ceph_mdsmap *m); >> > extern struct ceph_mdsmap *ceph_mdsmap_decode(void **p, void *end); >> > extern void ceph_mdsmap_destroy(struct ceph_mdsmap *m); >> > -- >> > 1.8.3.1 >> > >> > -- >> > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in >> > the body of a message to majordomo@xxxxxxxxxxxxxxx >> > More majordomo info at http://vger.kernel.org/majordomo-info.html >> -- >> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in >> the body of a message to majordomo@xxxxxxxxxxxxxxx >> More majordomo info at http://vger.kernel.org/majordomo-info.html > > I think we can't check mds presence by (mdsmap->m_max_mds == 0) > statement, because max_mds value does not strictly represent the number > of mds. For example, we have max_mds = 1 (by default) no matter have we > one mds or haven't. Is this a expected behavior? > I think so. mds may start later, so we should wait until timeout. nfs has the similar behavior. Regards Yan, Zheng -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
