diff options
| author | Alek P <[email protected]> | 2018-06-06 13:14:52 -0400 |
|---|---|---|
| committer | Brian Behlendorf <[email protected]> | 2018-06-06 10:14:52 -0700 |
| commit | 6969afcefdfb49fb9c0fcf56240e6eb133a2c4a8 (patch) | |
| tree | 6f8a0b0e0e24ab47ad7434b297367e42bbaf8b64 /cmd/zfs | |
| parent | 62841115bc77fbd8b55d89f07cbe1cc4fb3bc520 (diff) | |
Always continue recursive destroy after error
Currently, during a recursive zfs destroy the first error that is
encountered will stop the destruction of the datasets. Errors may
happen for a variety of reasons including competing deletions
and busy datasets.
This patch switches recursive destroy to always do a best-effort
recursive dataset destroy.
Reviewed-by: Brian Behlendorf <[email protected]>
Reviewed by: Matthew Ahrens <[email protected]>
Signed-off-by: Alek Pinchuk <[email protected]>
Closes #7574
Diffstat (limited to 'cmd/zfs')
| -rw-r--r-- | cmd/zfs/zfs_main.c | 11 |
1 files changed, 10 insertions, 1 deletions
diff --git a/cmd/zfs/zfs_main.c b/cmd/zfs/zfs_main.c index 7d81dcaee..b45e93c3f 100644 --- a/cmd/zfs/zfs_main.c +++ b/cmd/zfs/zfs_main.c @@ -1189,6 +1189,15 @@ destroy_callback(zfs_handle_t *zhp, void *data) zfs_unmount(zhp, NULL, cb->cb_force ? MS_FORCE : 0) != 0 || zfs_destroy(zhp, cb->cb_defer_destroy) != 0) { zfs_close(zhp); + /* + * When performing a recursive destroy we ignore errors + * so that the recursive destroy could continue + * destroying past problem datasets + */ + if (cb->cb_recurse) { + cb->cb_error = B_TRUE; + return (0); + } return (-1); } } @@ -1558,7 +1567,7 @@ zfs_do_destroy(int argc, char **argv) err = zfs_destroy_snaps_nvl(g_zfs, cb.cb_batchedsnaps, cb.cb_defer_destroy); } - if (err != 0) + if (err != 0 || cb.cb_error == B_TRUE) rv = 1; } |