Re: What is the best layer/device for a write-back cache based in nvram?

2016-09-14 Thread Thor Lancelot Simon
On Wed, Sep 14, 2016 at 03:16:13PM +, Eduardo Horvath wrote: > On Wed, 14 Sep 2016, Edgar Fu? wrote: > > > > 2- In scattered writes contained in a same slice, it allows to reduce > > > the number of writes. With RAID 5/6 there is a advantage, the parity > > > is written only one time for

Re: cgdstrategy: divide fault in supervisor mode

2016-09-14 Thread Alexander Nasonov
Michael van Elst wrote: > Right. This needs to be written differently. Instead of GETCGD_SOFTC() > use: > > cs = getcgd_softc(bp->b_dev); > if (!cs) { > bp->b_error = ENXIO; > biodone(bp); > return; > } I enabled DEBUG in the config and

Re: What is the best layer/device for a write-back cache based in nvram?

2016-09-14 Thread Hauke Fath
On Wed, 14 Sep 2016 13:54:34 +0200, Edgar Fuß wrote: >> 4- Faster synchronous writes. > Y E S. > This is the only point I fully aggree on. We've had severe problems with > brain-dead software (Firefox, Dropbox) performing tons of synchronous 4K > writes (on a bs=16K FFS) which nearly killed us

Re: What is the best layer/device for a write-back cache based in nvram?

2016-09-14 Thread Jose Luis Rodriguez Garcia
>> 1- There is no need to use parity map for the RAID 1/10/5/6. Usually >> the impact is small, but it can be noticeable in busy servers. >I don't notice it. When there is a crash, the time to rebuild the raid < 1min? ... >rather large. A segment should match a slice (or a number of them) >I

Re: What is the best layer/device for a write-back cache based in nvram?

2016-09-14 Thread Eduardo Horvath
On Wed, 14 Sep 2016, Edgar Fu? wrote: > > 2- In scattered writes contained in a same slice, it allows to reduce > > the number of writes. With RAID 5/6 there is a advantage, the parity > > is written only one time for several writes in the same slice, instead > > of one time for every write in

Re: What is the best layer/device for a write-back cache based in nvram?

2016-09-14 Thread Manuel Bouyer
On Wed, Sep 14, 2016 at 01:54:34PM +0200, Edgar Fuß wrote: > [...] > I would suppose LFS to perform great on a RAIDframe. Isn't Manuel Bouyer > using this in production? No, I played with LFS at some point but I never used it in production. -- Manuel Bouyer

Re: What is the best layer/device for a write-back cache based in nvram?

2016-09-14 Thread Edgar Fuß
I'm using a 12TB RAIDframe Level 5 RAID (4+1 discs) in production. There are 150 people's home directories and mail on FFFs file systems on it. > 1- There is no need to use parity map for the RAID 1/10/5/6. Usually > the impact is small, but it can be noticeable in busy servers. I don't notice

Re: cgdstrategy: divide fault in supervisor mode

2016-09-14 Thread Alexander Nasonov
Michael van Elst wrote: > Right. This needs to be written differently. Instead of GETCGD_SOFTC() > use: > > cs = getcgd_softc(bp->b_dev); > if (!cs) { > bp->b_error = ENXIO; > biodone(bp); > return; > } I tried something similar but with

Re: cgdstrategy: divide fault in supervisor mode

2016-09-14 Thread Michael van Elst
On Wed, Sep 14, 2016 at 07:18:56AM +0100, Alexander Nasonov wrote: > Michael van Elst wrote: > > Ah, maybe then: > > > > --- cgd.c 5 Aug 2016 08:24:46 - 1.110 > > +++ cgd.c 13 Sep 2016 21:43:27 - > > @@ -305,13 +305,17 @@ > > static void > > cgdstrategy(struct buf *bp)

Re: cgdstrategy: divide fault in supervisor mode

2016-09-14 Thread Alexander Nasonov
Michael van Elst wrote: > Ah, maybe then: > > --- cgd.c 5 Aug 2016 08:24:46 - 1.110 > +++ cgd.c 13 Sep 2016 21:43:27 - > @@ -305,13 +305,17 @@ > static void > cgdstrategy(struct buf *bp) > { > - struct cgd_softc *cs = getcgd_softc(bp->b_dev); > - struct