On Mon, Feb 1 at 16:12, Jake Carroll wrote:
Hi all.
Recent builds (b129, b130 and b131) have had me noticing some zpool
performance issues when scrubbing.
Running "bare" into some cheap SATA controllers, on a cheap mobo,
running 6GB of DDR2 + an Intel Q6600, with 4 * 1TB Samsung consumer
grade SATA drives, I've been accustomed to seeing around 150 to
180MB/sec scrubs on a single pool.
Until b129, 130 and 131 hit. I've got dedup=on, compression=on
(default, not gzip), no dedupe verify etc.
I think with dedupe, you've turned your scrub into a mostly random
operation.
I now see maybe 10MB/sec across 4 drives on scrub. Turning dedupe
off seemingly didn't help.
Disabling dedupe doesn't change the state of existing data. Unless
you've disabled dedupe, then re-copied all your data, I believe your
existing data is all still in the dedupe state.
Are we seeing an issue of dedupe here, or something less complex
entirely?
sounds like dedupe to me... My non-dedupe zpools are scrubbing at the
same rate as ever in b130 on multiple servers.
--eric
--
Eric D. Mudama
edmud...@mail.bounceswoosh.org
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
