diff options
author | delphij <delphij@FreeBSD.org> | 2014-05-09 07:04:12 +0000 |
---|---|---|
committer | delphij <delphij@FreeBSD.org> | 2014-05-09 07:04:12 +0000 |
commit | 0a785ddfd1f4133c792c2ca488647dcdd5ea6538 (patch) | |
tree | 46142fd7acd6e0a7e4dcd4c9418dbe79e5fa40bd /sys/cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c | |
parent | 0fce5e81cc639e45df5a18337cbee810899a3da5 (diff) | |
download | FreeBSD-src-0a785ddfd1f4133c792c2ca488647dcdd5ea6538.zip FreeBSD-src-0a785ddfd1f4133c792c2ca488647dcdd5ea6538.tar.gz |
MFC r264671 (MFV r264668):
4754 io issued to near-full luns even after setting noalloc threshold
4755 mg_alloc_failures is no longer needed
illumos/illumos@b6240e830b871f59c22a3918aebb3b36c872edba
Diffstat (limited to 'sys/cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c')
-rw-r--r-- | sys/cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c | 51 |
1 files changed, 6 insertions, 45 deletions
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c index 7f88d50..58aca2c 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c @@ -41,7 +41,7 @@ SYSCTL_NODE(_vfs_zfs, OID_AUTO, metaslab, CTLFLAG_RW, 0, "ZFS metaslab"); * avoid having to load lots of space_maps in a given txg. There are, * however, some cases where we want to avoid "fast" ganging and instead * we want to do an exhaustive search of all metaslabs on this device. - * Currently we don't allow any gang, zil, or dump device related allocations + * Currently we don't allow any gang, slog, or dump device related allocations * to "fast" gang. */ #define CAN_FASTGANG(flags) \ @@ -74,18 +74,6 @@ SYSCTL_INT(_vfs_zfs, OID_AUTO, condense_pct, CTLFLAG_RWTUN, " of in-memory counterpart"); /* - * This value defines the number of allowed allocation failures per vdev. - * If a device reaches this threshold in a given txg then we consider skipping - * allocations on that device. The value of zfs_mg_alloc_failures is computed - * in zio_init() unless it has been overridden in /etc/system. - */ -int zfs_mg_alloc_failures = 0; -TUNABLE_INT("vfs.zfs.mg_alloc_failures", &zfs_mg_alloc_failures); -SYSCTL_INT(_vfs_zfs, OID_AUTO, mg_alloc_failures, CTLFLAG_RWTUN, - &zfs_mg_alloc_failures, 0, - "Number of allowed allocation failures per vdev"); - -/* * The zfs_mg_noalloc_threshold defines which metaslab groups should * be eligible for allocation. The value is defined as a percentage of * a free space. Metaslab groups that have more free space than @@ -1707,10 +1695,7 @@ metaslab_sync_done(metaslab_t *msp, uint64_t txg) void metaslab_sync_reassess(metaslab_group_t *mg) { - int64_t failures = mg->mg_alloc_failures; - metaslab_group_alloc_update(mg); - atomic_add_64(&mg->mg_alloc_failures, -failures); /* * Preload the next potential metaslabs @@ -1737,7 +1722,7 @@ metaslab_distance(metaslab_t *msp, dva_t *dva) static uint64_t metaslab_group_alloc(metaslab_group_t *mg, uint64_t psize, uint64_t asize, - uint64_t txg, uint64_t min_distance, dva_t *dva, int d, int flags) + uint64_t txg, uint64_t min_distance, dva_t *dva, int d) { spa_t *spa = mg->mg_vd->vdev_spa; metaslab_t *msp = NULL; @@ -1764,10 +1749,9 @@ metaslab_group_alloc(metaslab_group_t *mg, uint64_t psize, uint64_t asize, spa_dbgmsg(spa, "%s: failed to meet weight " "requirement: vdev %llu, txg %llu, mg %p, " "msp %p, psize %llu, asize %llu, " - "failures %llu, weight %llu", - spa_name(spa), mg->mg_vd->vdev_id, txg, - mg, msp, psize, asize, - mg->mg_alloc_failures, msp->ms_weight); + "weight %llu", spa_name(spa), + mg->mg_vd->vdev_id, txg, + mg, msp, psize, asize, msp->ms_weight); mutex_exit(&mg->mg_lock); return (-1ULL); } @@ -1800,27 +1784,6 @@ metaslab_group_alloc(metaslab_group_t *mg, uint64_t psize, uint64_t asize, mutex_enter(&msp->ms_lock); /* - * If we've already reached the allowable number of failed - * allocation attempts on this metaslab group then we - * consider skipping it. We skip it only if we're allowed - * to "fast" gang, the physical size is larger than - * a gang block, and we're attempting to allocate from - * the primary metaslab. - */ - if (mg->mg_alloc_failures > zfs_mg_alloc_failures && - CAN_FASTGANG(flags) && psize > SPA_GANGBLOCKSIZE && - activation_weight == METASLAB_WEIGHT_PRIMARY) { - spa_dbgmsg(spa, "%s: skipping metaslab group: " - "vdev %llu, txg %llu, mg %p, msp[%llu] %p, " - "psize %llu, asize %llu, failures %llu", - spa_name(spa), mg->mg_vd->vdev_id, txg, mg, - msp->ms_id, msp, psize, asize, - mg->mg_alloc_failures); - mutex_exit(&msp->ms_lock); - return (-1ULL); - } - - /* * Ensure that the metaslab we have selected is still * capable of handling our request. It's possible that * another thread may have changed the weight while we @@ -1859,8 +1822,6 @@ metaslab_group_alloc(metaslab_group_t *mg, uint64_t psize, uint64_t asize, if ((offset = metaslab_block_alloc(msp, asize)) != -1ULL) break; - atomic_inc_64(&mg->mg_alloc_failures); - metaslab_passivate(msp, metaslab_block_maxsize(msp)); mutex_exit(&msp->ms_lock); } @@ -2015,7 +1976,7 @@ top: ASSERT(P2PHASE(asize, 1ULL << vd->vdev_ashift) == 0); offset = metaslab_group_alloc(mg, psize, asize, txg, distance, - dva, d, flags); + dva, d); if (offset != -1ULL) { /* * If we've just selected this metaslab group, |