Date: Thu, 6 Sep 2018 01:11:30 -0700 From: Mark Millard <marklmi@yahoo.com> To: Kirk McKusick <mckusick@mckusick.com> Cc: FreeBSD Filesystems <freebsd-fs@FreeBSD.org>, bob prohaska <fbsd@www.zefox.net> Subject: Re: CFT: TRIM Consolodation on UFS/FFS filesystems Message-ID: <97765759-B81A-44F0-98A0-F7B9D05F7432@yahoo.com> In-Reply-To: <20180906001217.GB818@www.zefox.net> References: <A751682B-6718-4E9F-87DB-2C7AA7076854@yahoo.com> <201809052207.w85M7GS2000773@chez.mckusick.com> <20180906001217.GB818@www.zefox.net>
index | next in thread | previous in thread | raw e-mail
[Just correcting a bad claim about the timing of the two buildworld runs that were referenced. It could be misleading to Kirk for his requested information.] On 2018-Sep-5, at 5:12 PM, bob prohaska <fbsd at www.zefox.net> wrote: > On Wed, Sep 05, 2018 at 03:07:16PM -0700, Kirk McKusick wrote: >> >> It is a bit tricky to just turn off TRIM and then measure it, because >> in the immediate aftermath of its having been used, it will leave >> behind a legacy of easier to use flash areas yet not have the cost >> of keeping them cleaned up. So, it might initially look like enabling >> TRIM is a bad idea. Thus you would have to run many installworlds >> without TRIM enabled to see what the long-term result of not using >> TRIM turns out to be. >> > > Just for fun I ran a (somewhat absurd) -j4 buildworld on RPI3 using 6 GB of swap, > three on USB and three on microSD, just to see if anything interesting (bad) happened. > The process took about 24 hours, and the oversuppy of swap didn't cause any obvious > problems. Next I turned on TRIM and re-ran the buildworld script. > > There were no obvious problems, but the process took about an extra hour. buildworld did not take an hour longer for one vs. the other based on the timestamps in the log files: trim off: World build started on Sun Sep 2 20:28:12 PDT 2018 . . . World build completed on Mon Sep 3 21:35:47 PDT 2018 So somewhat over 25 hours 7 minutes. trim on: World build started on Tue Sep 4 00:02:36 PDT 2018 . . . World build completed on Wed Sep 5 01:12:47 PDT 2018 So somewhat over 25 hours 10 minutes. I get an under 5 minute difference from those timestamps. > Since /var, /tmp and /usr were all on USB there was no hope TRIM could be > any help on the busy filesystems. TRIM was enabled on microSD, but it had > little to do. There does seem to be a modest penalty for using TRIM when > it can't help much. > > Is there any hope of implementing something like TRIM for USB on the Pi? > It appears that congestion on USB is a serious bottlneck from time to time > just for traffic with /tmp and /usr. Adding swap to the mix makes it worse. > > Log files are at > http://www.zefox.net/~fbsd/rpi3/swaptests/r338342/3gbsd_3gbusb/ > in case they're of interest. === Mark Millard marklmi at yahoo.com ( dsl-only.net went away in early 2018-Mar)help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?97765759-B81A-44F0-98A0-F7B9D05F7432>