On Tue, Dec 11, 2018 at 9:56 AM Dongsheng Yang
<dongsheng.yang@xxxxxxxxxxxx> wrote:
>
> Introduce a new option abort_on_full, default to false. Then
> we can get -ENOSPC when the pool is full, or reaches quota.
>
> Signed-off-by: Dongsheng Yang <dongsheng.yang@xxxxxxxxxxxx>
> ---
> drivers/block/rbd.c | 10 ++++++++++
> 1 file changed, 10 insertions(+)
>
> diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c
> index 8e5140b..7cd35e2 100644
> --- a/drivers/block/rbd.c
> +++ b/drivers/block/rbd.c
> @@ -744,6 +744,7 @@ enum {
> Opt_lock_on_read,
> Opt_exclusive,
> Opt_notrim,
> + Opt_abort_on_full,
> Opt_err
> };
>
> @@ -760,6 +761,7 @@ enum {
> {Opt_lock_on_read, "lock_on_read"},
> {Opt_exclusive, "exclusive"},
> {Opt_notrim, "notrim"},
> + {Opt_abort_on_full, "abort_on_full"},
> {Opt_err, NULL}
> };
>
> @@ -770,6 +772,7 @@ struct rbd_options {
> bool lock_on_read;
> bool exclusive;
> bool trim;
> + bool abort_on_full;
> };
>
> #define RBD_QUEUE_DEPTH_DEFAULT BLKDEV_MAX_RQ
> @@ -778,6 +781,7 @@ struct rbd_options {
> #define RBD_LOCK_ON_READ_DEFAULT false
> #define RBD_EXCLUSIVE_DEFAULT false
> #define RBD_TRIM_DEFAULT true
> +#define RBD_ABORT_ON_FULL_DEFAULT false
>
> struct parse_rbd_opts_ctx {
> struct rbd_spec *spec;
> @@ -841,6 +845,9 @@ static int parse_rbd_opts_token(char *c, void *private)
> case Opt_notrim:
> pctx->opts->trim = false;
> break;
> + case Opt_abort_on_full:
> + pctx->opts->abort_on_full = true;
> + break;
> default:
> /* libceph prints "bad option" msg */
> return -EINVAL;
> @@ -5393,6 +5400,7 @@ static int rbd_add_parse_args(const char *buf,
> pctx.opts->lock_on_read = RBD_LOCK_ON_READ_DEFAULT;
> pctx.opts->exclusive = RBD_EXCLUSIVE_DEFAULT;
> pctx.opts->trim = RBD_TRIM_DEFAULT;
> + pctx.opts->abort_on_full = RBD_ABORT_ON_FULL_DEFAULT;
>
> copts = ceph_parse_options(options, mon_addrs,
> mon_addrs + mon_addrs_size - 1,
> @@ -5851,6 +5859,8 @@ static ssize_t do_rbd_add(struct bus_type *bus,
> goto err_out_args;
> }
>
> + rbdc->client->osdc.abort_on_full = rbd_opts->abort_on_full;
> +
> /* pick the pool */
> rc = ceph_pg_poolid_by_name(rbdc->client->osdc.osdmap, spec->pool_name);
> if (rc < 0) {
Hi Dongsheng,
I think it should be a libceph option, not rbd. By default, libceph
instances are shared and we don't want someone mapping a new image with
abort_on_full to interfere with existing mappings.
Thanks,
Ilya
