Date: Thu, 18 Jul 2019 08:04:20 -0700 From: bob prohaska <fbsd@www.zefox.net> To: Robert Crowston <crowston@protonmail.com> Cc: freebsd-arm@freebsd.org, bob prohaska <fbsd@www.zefox.net> Subject: Re: Lethargic rpi3, but seemingly still running Message-ID: <20190718150420.GA4325@www.zefox.net> In-Reply-To: <C6eXqU7tLQ8FRlqfpu5X-rN371auRUle8HdhWkDLP8xvdp3K7ImuV6v0dpCezw_UUdslxKBfYpyyYhzAcMuGbNIaEqfwqPbpQFC3V-9vhzE=@protonmail.com> References: <20190718034838.GA1921@www.zefox.net> <C6eXqU7tLQ8FRlqfpu5X-rN371auRUle8HdhWkDLP8xvdp3K7ImuV6v0dpCezw_UUdslxKBfYpyyYhzAcMuGbNIaEqfwqPbpQFC3V-9vhzE=@protonmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Jul 18, 2019 at 06:29:37AM +0000, Robert Crowston wrote: > If you compiled -CURRENT with the default options it will have a bunch of debug enabled including witness. Are you spending a long time in syscalls? > Not according to top, it's 99.8% idle. Is there a better way to measure? The system _is_ default, but that isn't new. Firefox has built before and didn't display behavior like this. Buildworld has been taking about 25% longer than usual in the past month or so, but but it hasn't behaved so badly as this. Buildworld also doesn't invoke bsdtar; I wonder if it's the culprit. Allowed to run overnight it came unstuck and is now running at near full tilt compiling LLVM. Thanks for writing! bob prohaska > On Thu, Jul 18, 2019 at 04:48, bob prohaska <fbsd@www.zefox.net> wrote: > > > I'm playing with an RPI3 running -current at r349989 and trying to build > > a port (firefox) with the ports tree at (....long delay to svnlite info > > command.....) 506825 using portmaster under the portmaster-devel script. > > > > The system is still running and mostly idle, as reported by top: > > > > last pid: 8739; load averages: 0.00, 0.02, 0.00 up 1+12:03:29 20:29:20 > > 36 processes: 1 running, 35 sleeping > > CPU: 0.0% user, 0.0% nice, 0.1% system, 0.1% interrupt, 99.8% idle > > Mem: 22M Active, 720M Inact, 2324K Laundry, 137M Wired, 57M Buf, 27M Free > > Swap: 6349M Total, 23M Used, 6326M Free > > > > PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAN > > 8374 root 1 26 0 27M 9160K biord 1 1:21 0.00% bsdtar > > 8739 root 1 21 0 21M 8108K sbwait 3 0:00 0.00% sshd > > 21314 bob 1 20 0 14M 1120K CPU2 2 3:44 0.20% top > > 21512 bob 1 21 0 13M 1108K ttyin 1 0:00 0.00% tcsh > > 970 root 1 20 0 16M 848K select 0 0:09 0.02% sendma > > 957 root 1 21 0 20M 688K select 0 0:05 0.00% sshd > > 38012 root 1 52 0 15M 656K wait 2 0:58 0.00% bash > > 973 smmsp 1 20 0 16M 548K pause 3 0:00 0.00% sendma > > 2036 bob 1 20 0 22M 544K select 1 0:22 0.02% sshd > > 977 root 1 20 0 12M 412K nanslp 0 0:03 0.00% cron > > 21511 bob 1 20 0 22M 408K select 0 0:00 0.00% sshd > > 826 root 1 20 0 11M 404K select 3 0:03 0.00% syslog > > 2001 bob 1 20 0 22M 244K select 1 0:04 0.00% sshd > > 2007 root 1 20 0 14M 140K pause 2 0:00 0.00% csh > > > > Every once in a while bsdtar will use a fraction of a percent of cpu, but > > that seems to be all that's happening. The timestamp updates every other > > second, typing echoes as expected but something like pwd takes twenty > > seconds to answer. The controlling terminal for portmaster-devel has been > > sitting at > > > > => SHA256 Checksum OK for rust/2019-05-23/cargo-0.36.0-aarch64-unknown-freebsd.tar.gz. > > for about two hours. > > > > There are sporadic reports of > > swap_pager: indefinite wait buffer: bufobj: 0, blkno: 516, size: 4096 > > on the console, but they're sparse. At least according to top there's > > no significant swap usage. > > > > This behavior isn't entirely new, but this is the first time it's been > > bad enough to seem worth reporting. The filesystem is 74% full, so it's > > not out of storage: > > Filesystem 1K-blocks Used Avail Capacity Mounted on > > /dev/ufs/rootfs 116766684 78989800 28435552 74% / > > devfs 1 1 0 100% /dev > > /dev/msdosfs/MSDOSBOOT 51096 13552 37544 27% /boot/msdos > > > > Meanwhile an RPI2 running 11.3-BETA1 #0 r348441 is happily grinding away > > on the same problem with ports at 506786. It answered svnlite info /usr/ports > > in less than a second. > > > > Thanks for reading, and any thoughts. > > > > bob prohaska > > > > _______________________________________________ > > freebsd-arm@freebsd.org mailing list > > https://lists.freebsd.org/mailman/listinfo/freebsd-arm > > To unsubscribe, send any mail to "freebsd-arm-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20190718150420.GA4325>