diff options
| author | Shaohua Li <shli@fb.com> | 2017-02-16 04:58:05 +0100 |
|---|---|---|
| committer | Shaohua Li <shli@fb.com> | 2017-03-17 00:55:51 +0100 |
| commit | 84890c03b6c5d7e8d76ea5e20b6aaf7e7ad410f0 (patch) | |
| tree | 68faa6589a2f76235444403cbef5ebb7d55ad157 /drivers/md/raid5-cache.c | |
| parent | md/raid5: prioritize stripes for writeback (diff) | |
| download | linux-84890c03b6c5d7e8d76ea5e20b6aaf7e7ad410f0.tar.xz linux-84890c03b6c5d7e8d76ea5e20b6aaf7e7ad410f0.zip | |
md/raid5-cache: bump flush stripe batch size
Bump the flush stripe batch size to 2048. For my 12 disks raid
array, the stripes takes:
12 * 4k * 2048 = 96MB
This is still quite small. A hardware raid card generally has 1GB size,
which we suggest the raid5-cache has similar cache size.
The advantage of a big batch size is we can dispatch a lot of IO in the
same time, then we can do some scheduling to make better IO pattern.
Last patch prioritizes stripes, so we don't worry about a big flush
stripe batch will starve normal stripes.
Signed-off-by: Shaohua Li <shli@fb.com>
Diffstat (limited to '')
| -rw-r--r-- | drivers/md/raid5-cache.c | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/drivers/md/raid5-cache.c b/drivers/md/raid5-cache.c index 3f307be01b10..738e180d3740 100644 --- a/drivers/md/raid5-cache.c +++ b/drivers/md/raid5-cache.c @@ -43,7 +43,7 @@ /* wake up reclaim thread periodically */ #define R5C_RECLAIM_WAKEUP_INTERVAL (30 * HZ) /* start flush with these full stripes */ -#define R5C_FULL_STRIPE_FLUSH_BATCH 256 +#define R5C_FULL_STRIPE_FLUSH_BATCH(conf) (conf->max_nr_stripes / 4) /* reclaim stripes in groups */ #define R5C_RECLAIM_STRIPE_GROUP (NR_STRIPE_HASH_LOCKS * 2) @@ -381,7 +381,7 @@ void r5c_check_cached_full_stripe(struct r5conf *conf) * or a full stripe (chunk size / 4k stripes). */ if (atomic_read(&conf->r5c_cached_full_stripes) >= - min(R5C_FULL_STRIPE_FLUSH_BATCH, + min(R5C_FULL_STRIPE_FLUSH_BATCH(conf), conf->chunk_sectors >> STRIPE_SHIFT)) r5l_wake_reclaim(conf->log, 0); } @@ -1393,7 +1393,7 @@ static void r5c_do_reclaim(struct r5conf *conf) stripes_to_flush = R5C_RECLAIM_STRIPE_GROUP; else if (total_cached > conf->min_nr_stripes * 1 / 2 || atomic_read(&conf->r5c_cached_full_stripes) - flushing_full > - R5C_FULL_STRIPE_FLUSH_BATCH) + R5C_FULL_STRIPE_FLUSH_BATCH(conf)) /* * if stripe cache pressure moderate, or if there is many full * stripes,flush all full stripes |