Date: Thu, 21 Dec 2023 19:56:21 +0100 (CET) From: Ronald Klop <ronald-lists@klop.ws> To: void <void@f-m.fm> Cc: freebsd-fs@freebsd.org Subject: Re: measuring swap partition speed Message-ID: <1439231787.9333.1703184981914@localhost> In-Reply-To: <ZYRe3GFrf1fSIPjk@int21h> References: <D9C150E2-8453-463B-9A23-B0474D0AE390.ref@yahoo.com> <D9C150E2-8453-463B-9A23-B0474D0AE390@yahoo.com> <ZYRe3GFrf1fSIPjk@int21h>
next in thread | previous in thread | raw e-mail | index | archive | help
------=_Part_9332_1497520887.1703184981868 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit A bit weird that a simple dd is so slow. Just a quick thought. Are your partitions aligned properly? Is other IO going on on the same time? What does gstat say about %util, queue length and all the other stats while running the dd? Or "iostat -x -d 1". Could you try if another disk has the same issues? How is your disk connected? USB-to-SATA-adapter? Any output of dmesg? Regards, Ronald. Van: void <void@f-m.fm> Datum: donderdag, 21 december 2023 16:50 Aan: freebsd-fs@freebsd.org Onderwerp: Re: measuring swap partition speed > > On Wed, Dec 20, 2023 at 07:48:14PM -0800, Mark Millard wrote: > > ># swapoff /dev/label/growfs_swap > ># dd if=/dev/urandom of=/dev/da0s2b bs=8k count=250000 conv=sync status=progress > >^C478830592 bytes (479 MB, 457 MiB) transferred 22.001s, 22 MB/s > >60557+0 records in > >60556+0 records out > >496074752 bytes transferred in 22.790754 secs (21766491 bytes/sec) > > 22MB/s is usable, I think. In my context, I'd be satisfied with that. > My context differs from yours slightly in that yours is SSD and mine > is spinning rust. > > This is unusable: > # dd if=/dev/urandom of=/dev/da0p4 bs=8k count=250000 conv=sync status=progress > ^C11862016 bytes (12 MB, 11 MiB) transferred 40.063s, 296 kB/s > > because it's way too slow. Swap never gets fully reclaimed, > thrashing happens, loads of other followon effects happen. > The same partition formatted as ufs reports 113 MB/s. Multiple swap partitions > have been tested, then converted to ufs. Results are the same. > > There are no reported errors in smartctl. Long smartctl tests run monthly. > > 5 Reallocated_Sector_Ct PO--CK 100 100 050 - 0 > 9 Power_On_Hours -O--CK 001 001 000 - 48992 > 196 Reallocated_Event_Count -O--CK 100 100 000 - 0 > 197 Current_Pending_Sector -O--CK 100 100 000 - 0 > 198 Offline_Uncorrectable ----CK 100 100 000 - 0 > > I can't find any hardware problem here. Possible workarounds, bearing in mind I'm not versant in C so it's not like I can fix this myself in code: > > 1. swap as swapfile and not partition [a] > 2. swap as nfs [b] > 3. swapoff & swapon script running every minute [c] > 4. just turn all swap off and reboot after crashing (undesirable) > 5. use another OS that doesn't have this problem > > [a] not tried yet, and i hope it works. Legacy info suggests swap as partition is usually > faster than filesystem-based swap. But the reverse might be the case here. > > [b] also not tried. This, I imagine, would be filesystem only (I'm unsure a zfs volume can > be exported to look like a mountable partition to the client) > > [c] https://github.com/Freaky/swapflush.git - usually works but maybe i need to run it every minute instead of every five mins. For testing, this script was disabled. > > Any additional suggestions on how to overcome this problem gratefully received. > > -- > > > > ------=_Part_9332_1497520887.1703184981868 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit <html><head></head><body><br> A bit weird that a simple dd is so slow.<br> <br> Just a quick thought.<br> Are your partitions aligned properly?<br> Is other IO going on on the same time?<br> What does gstat say about %util, queue length and all the other stats while running the dd? Or "iostat -x -d 1".<br> <br> Could you try if another disk has the same issues?<br> <br> How is your disk connected? USB-to-SATA-adapter? Any output of dmesg?<br> <br> Regards,<br> Ronald.<br> <br> <p><strong>Van:</strong> void <void@f-m.fm><br> <strong>Datum:</strong> donderdag, 21 december 2023 16:50<br> <strong>Aan:</strong> freebsd-fs@freebsd.org<br> <strong>Onderwerp:</strong> Re: measuring swap partition speed</p> <blockquote style="padding-right: 0px; padding-left: 5px; margin-left: 5px; border-left: #000000 2px solid; margin-right: 0px"> <div class="MessageRFC822Viewer" id="P"> <div class="TextPlainViewer" id="P.P">On Wed, Dec 20, 2023 at 07:48:14PM -0800, Mark Millard wrote:<br> <br> ># swapoff /dev/label/growfs_swap<br> ># dd if=/dev/urandom of=/dev/da0s2b bs=8k count=250000 conv=sync status=progress<br> >^C478830592 bytes (479 MB, 457 MiB) transferred 22.001s, 22 MB/s<br> >60557+0 records in<br> >60556+0 records out<br> >496074752 bytes transferred in 22.790754 secs (21766491 bytes/sec)<br> <br> 22MB/s is usable, I think. In my context, I'd be satisfied with that.<br> My context differs from yours slightly in that yours is SSD and mine<br> is spinning rust.<br> <br> This is unusable:<br> # dd if=/dev/urandom of=/dev/da0p4 bs=8k count=250000 conv=sync status=progress<br> ^C11862016 bytes (12 MB, 11 MiB) transferred 40.063s, 296 kB/s <br> <br> because it's way too slow. Swap never gets fully reclaimed,<br> thrashing happens, loads of other followon effects happen.<br> The same partition formatted as ufs reports 113 MB/s. Multiple swap partitions<br> have been tested, then converted to ufs. Results are the same.<br> <br> There are no reported errors in smartctl. Long smartctl tests run monthly.<br> <br> 5 Reallocated_Sector_Ct PO--CK 100 100 050 - 0<br> 9 Power_On_Hours -O--CK 001 001 000 - 48992<br> 196 Reallocated_Event_Count -O--CK 100 100 000 - 0<br> 197 Current_Pending_Sector -O--CK 100 100 000 - 0<br> 198 Offline_Uncorrectable ----CK 100 100 000 - 0<br> <br> I can't find any hardware problem here. Possible workarounds, bearing in mind I'm not versant in C so it's not like I can fix this myself in code:<br> <br> 1. swap as swapfile and not partition [a]<br> 2. swap as nfs [b]<br> 3. swapoff & swapon script running every minute [c]<br> 4. just turn all swap off and reboot after crashing (undesirable)<br> 5. use another OS that doesn't have this problem<br> <br> [a] not tried yet, and i hope it works. Legacy info suggests swap as partition is usually<br> faster than filesystem-based swap. But the reverse might be the case here.<br> <br> [b] also not tried. This, I imagine, would be filesystem only (I'm unsure a zfs volume can<br> be exported to look like a mountable partition to the client)<br> <br> [c] <a href="https://github.com/Freaky/swapflush.git">https://github.com/Freaky/swapflush.git</a>; - usually works but maybe i need to run it every minute instead of every five mins. For testing, this script was disabled.<br> <br> Any additional suggestions on how to overcome this problem gratefully received.<br> <br> -- <br> </div> <hr></div> </blockquote> <br> </body></html> ------=_Part_9332_1497520887.1703184981868--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1439231787.9333.1703184981914>