31 Jan
2018
31 Jan
'18
8:24 a.m.
On Wed, January 31, 2018 9:03 am, Robert Bielik wrote:
Ok ð although this is jack2 built *without* dbus
and running on a
headless RPi3.
How does the rc.local start jackd under the user account? Maybe it could
start a screen session at that point, and when you ssh in you can connect
to the screen session. If you are not familiar with screen it is an
application that (among other features) will let you disconnect an ssh
session while the underlying terminal session continues to run (usually
breaking the ssh session will kill the terminal session associated with
that login). When you connect again with ssh you start screen with the -r
argument to restore connection to the already running screen session
rather than starting a new session.
--
Chris Caudle
31 Jan
31 Jan
8:30 a.m.
New subject: [Jack-Devel] Usage feasibility Q
How does the rc.local start jackd under the user
account? Maybe it could
start a screen session at that point, and when you ssh in you can connect
to the screen session. If you are not familiar with screen it is an
application that (among other features) will let you disconnect an ssh
session while the underlying terminal session continues to run (usually
breaking the ssh session will kill the terminal session associated with
that login). When you connect again with ssh you start screen with the -r
argument to restore connection to the already running screen session
rather than starting a new session.
As shown here (I wrote that msg in the forum):
http://forum.audioinjector.net/viewtopic.php?f=5&t=2727&e=1&vie…
sudo runuser -u pi -- /home/pi/start_jack
And right now everything works, strace gives me:
...
connect(3, {sa_family=AF_UNIX, sun_path="/dev/shm/jack_default_1000_0"}, 110) =
0
...
And jack_lsp works as expected. Contents in /dev/shm:
pi@realtimepi:~ $ ls -al /dev/shm/
total 69804
drwxrwxrwt 2 root root 160 Jan 31 15:24 .
drwxr-xr-x 14 root root 3300 Jan 31 15:19 ..
-rw-r--r-- 1 pi pi 71390368 Jan 31 15:22 jack-1000-0
-rw-r--r-- 1 pi pi 1240 Jan 31 15:22 jack-1000-1
srwxr-xr-x 1 pi pi 0 Jan 31 15:22 jack_default_1000_0
-rwxr-xr-x 1 pi pi 12 Jan 31 15:22 jack_sem.1000_default_freewheel
-rwxr-xr-x 1 pi pi 12 Jan 31 15:22 jack_sem.1000_default_system
-rw-rw-rw- 1 pi pi 70744 Jan 31 15:22 jack-shm-registry
When it is not working, the "jack_default_1000_0" file does not exist anymore,
but the jackd server is running.
I'll lookup the screen option, although this feels like more fundametal problem. The
files in /dev/shm aren't supposed to vanish during lifetime of jackd process, right ?
Regards
/Robert
8:35 a.m.
New subject: [Jack-Devel] Usage feasibility Q
I'll lookup the screen option, although this feels
like more fundametal
problem. The files in /dev/shm aren't supposed to vanish during lifetime of
jackd process, right ?
Ok, when logging out of all ssh sessions, the contents of /dev/shm seems to vanish, at
least it cannot be seen in the newly attached ssh session.
But df -h shows that the jackd process still holds the 69MB it had allocated:
pi@realtimepi:~ $ df -h
Filesystem Size Used Avail Use% Mounted on
/dev/root 29G 1.6G 26G 6% /
devtmpfs 459M 0 459M 0% /dev
tmpfs 160M 69M 92M 43% /dev/shm <<<<<<<<
tmpfs 464M 12M 452M 3% /run
tmpfs 5.0M 4.0K 5.0M 1% /run/lock
tmpfs 464M 0 464M 0% /sys/fs/cgroup
/dev/mmcblk0p1 41M 35M 6.8M 84% /boot
tmpfs 93M 0 93M 0% /run/user/1000
Regards
/Robert
8:58 a.m.
New subject: [Jack-Devel] ?==?utf-8?q? Usage feasibility Q
BTW, do a:
$ lsof -p `pgrep jackdbus ` | grep log
If you are lucky your jackd logs to a logfile and you might be able to find out what
happened.
Cheers, RalfD
9:07 a.m.
New subject: [Jack-Devel] Usage feasibility Q
$ lsof -p `pgrep jackdbus ` | grep log
Output of lsof -p `pgrep jackd` (I don't have jackdbus):
pi@realtimepi:~ $ lsof -p `pgrep jackd`
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
jackd 578 pi cwd DIR 179,2 4096 2 /
jackd 578 pi rtd DIR 179,2 4096 2 /
jackd 578 pi txt REG 179,2 34080 42764 /usr/local/bin/jackd
jackd 578 pi mem REG 179,2 168156 254843 /usr/local/lib/jack/jack_alsa.so
jackd 578 pi DEL REG 0,16 11508 /dev/shm/jack-1000-0
jackd 578 pi mem REG 179,2 902752 10550
/usr/lib/arm-linux-gnueabihf/libasound.so.2.0.0
jackd 578 pi DEL REG 0,16 11507 /dev/shm/jack-shm-registry
jackd 578 pi mem REG 179,2 1234700 1887
/lib/arm-linux-gnueabihf/libc-2.24.so
jackd 578 pi mem REG 179,2 116372 1902
/lib/arm-linux-gnueabihf/libgcc_s.so.1
jackd 578 pi mem REG 179,2 452152 1914
/lib/arm-linux-gnueabihf/libm-2.24.so
jackd 578 pi mem REG 179,2 1269620 10687
/usr/lib/arm-linux-gnueabihf/libstdc++.so.6.0.22
jackd 578 pi mem REG 179,2 9800 1896
/lib/arm-linux-gnueabihf/libdl-2.24.so
jackd 578 pi mem REG 179,2 26632 1952
/lib/arm-linux-gnueabihf/librt-2.24.so
jackd 578 pi mem REG 179,2 127300 1948
/lib/arm-linux-gnueabihf/libpthread-2.24.so
jackd 578 pi mem REG 179,2 1241196 40853
/usr/local/lib/libjackserver.so.0.1.0
jackd 578 pi mem REG 179,2 21868 10548
/usr/lib/arm-linux-gnueabihf/libarmmem.so
jackd 578 pi mem REG 179,2 138576 1840
/lib/arm-linux-gnueabihf/ld-2.24.so
jackd 578 pi DEL REG 0,16 8034
/dev/shm/jack_sem.1000_default_freewheel
jackd 578 pi mem CHR 116,16 11413 /dev/snd/pcmC0D0p
jackd 578 pi mem CHR 116,24 11414 /dev/snd/pcmC0D0c
jackd 578 pi DEL REG 0,16 11510
/dev/shm/jack_sem.1000_default_system
jackd 578 pi DEL REG 0,16 11509 /dev/shm/jack-1000-1
jackd 578 pi 0r CHR 1,3 0t0 1028 /dev/null
jackd 578 pi 1w CHR 5,1 0t0 1035 /dev/console
jackd 578 pi 2u unix 0xb6447480 0t0 10734 type=STREAM
jackd 578 pi 3u REG 0,16 12 11510
/dev/shm/jack_sem.1000_default_system (deleted)
jackd 578 pi 4u CHR 116,0 0t0 11412 /dev/snd/controlC0
jackd 578 pi 5u unix 0xb78a4380 0t0 8031 /dev/shm/jack_default_1000_0
type=STREAM
jackd 578 pi 6u CHR 116,16 0t0 11413 /dev/snd/pcmC0D0p
jackd 578 pi 7u CHR 116,24 0t0 11414 /dev/snd/pcmC0D0c
jackd 578 pi 8u unix 0xb78a7b80 0t0 8033 type=STREAM
jackd 578 pi 9u REG 0,16 12 8034
/dev/shm/jack_sem.1000_default_freewheel (deleted)
jackd 578 pi 10u unix 0xb78a7800 0t0 11511 /dev/shm/jack_default_1000_0
type=STREAM
Interesting to see the DEL for file descriptor.
If you are lucky your jackd logs to a logfile and you
might be able to find out
what happened.
I'll search for a log file.
Regards
/Robert
9:19 a.m.
New subject: [Jack-Devel] ?==?utf-8?q? Usage feasibility Q
Am Mittwoch, 31. Januar 2018 17:07 CET, Robert Bielik <Robert.Bielik(a)dirac.com>
schrieb:
$ lsof -p
`pgrep jackdbus ` | grep log
Output of lsof -p `pgrep jackd` (I don't have jackdbus):
pi@realtimepi:~ $ lsof -p `pgrep jackd`
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
jackd 578 pi cwd DIR 179,2 4096 2 /
[...]
jackd 578 pi 10u unix 0xb78a7800 0t0 11511 /dev/shm/jack_default_1000_0
type=STREAM
Interesting to see the DEL for file descriptor.
I'll search for a log file.
That's what the '| grep log' was meant for ;-)
Looks like you don't have one. Maybe you can append '> /tmp/jackd.log
2>&1 ' to your jackd invocation
in your '/home/pi/start_jack' script?
Cheers, RalfD
Regards
/Robert
_______________________________________________
Jack-Devel mailing list
Jack-Devel(a)lists.jackaudio.org
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
8:33 a.m.
New subject: [Jack-Devel] ?==?utf-8?q? Usage feasibility Q
Am Mittwoch, 31. Januar 2018 16:24 CET, "Chris Caudle"
<chris(a)chriscaudle.org> schrieb:
On Wed, January 31, 2018 9:03 am, Robert Bielik
wrote:
But the OP clearly stated that it worked during a ssh login. It's only later
on that it stops working.
Cheers, RalfD
Ok 😊 although this is jack2 built *without*
dbus and running on a
headless RPi3.
How does the rc.local start jackd under the user account? Maybe it could
start a screen session at that point, and when you ssh in you can connect
to the screen session.
8:47 a.m.
New subject: [Jack-Devel] Usage feasibility Q
On Wed, January 31, 2018 9:33 am, Ralf Mattes wrote:
But the OP clearly stated that it worked during a ssh
login. It's only
later on that it stops working.
It works during the first ssh login, but not during subsequent logins.
Screen would avoid that issue by never really closing down the user
session.
--
Chris Caudle
8:51 a.m.
New subject: [Jack-Devel] ?==?utf-8?q? Usage feasibility Q
Am Mittwoch, 31. Januar 2018 16:47 CET, "Chris Caudle"
<chris(a)chriscaudle.org> schrieb:
On Wed, January 31, 2018 9:33 am, Ralf Mattes wrote:
Only if the assumption that the logout triggers socket removal is true (it might
as well be something that happens after a certain time etc.).
And even then I might be a "smarter idea" to fix that buggy behavior instead of
duct-tapeing the problem.
Cheers, RalfD
But the OP clearly stated that it worked during a
ssh login. It's only
later on that it stops working.
It works during the first ssh login, but not during subsequent logins.
Screen would avoid that issue by never really closing down the user
session.
--
Chris Caudle
_______________________________________________
Jack-Devel mailing list
Jack-Devel(a)lists.jackaudio.org
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
1:15 p.m.
New subject: [Jack-Devel] Usage feasibility Q
"Chris Caudle" <chris(a)chriscaudle.org> writes:
On Wed, January 31, 2018 9:33 am, Ralf Mattes wrote:
Huh. Perhaps start jackd with nohup ? Maybe it reacts to a hangup
signal in some manner?
--
David Kastrup
But the OP clearly stated that it worked during a
ssh login. It's only
later on that it stops working.
It works during the first ssh login, but not during subsequent logins.
Screen would avoid that issue by never really closing down the user
session.
1:19 p.m.
New subject: [Jack-Devel] Usage feasibility Q
David Kastrup <dak(a)gnu.org> writes:
"Chris Caudle" <chris(a)chriscaudle.org>
writes:
Never mind. Saw in another answer that systemd's idea of ending a user
session is at fault. IIRC, systemd even overrides an explicit nohup
anyway (WTF?!?).
--
David Kastrup
On Wed, January 31, 2018 9:33 am, Ralf Mattes
wrote:
Huh. Perhaps start jackd with nohup ? Maybe it reacts to a hangup
signal in some manner? But the OP clearly stated that it worked during a
ssh login. It's only
later on that it stops working.
It works during the first ssh login, but not during subsequent logins.
Screen would avoid that issue by never really closing down the user
session.
1:46 p.m.
New subject: [Jack-Devel] Usage feasibility Q
On Wed, January 31, 2018 2:19 pm, David Kastrup wrote:
Never mind. Saw in another answer that systemd's
idea of ending a user
session is at fault. IIRC, systemd even overrides an explicit nohup
anyway (WTF?!?).
I had forgotten, but after seeing the answer I remember when this came up
a couple of years ago, systemd proposed this behavior as a security
feature, some subset of users yelled about how it would break their usual
work flow that involved using nohup to keep jobs running in the background
when they logged out, so the behavior was made configurable but "secure by
default." Obviously not many people make use of the configurability,
because when this comes up it takes a pretty deep google search to stumble
on the answer, seems hardly anyone remembers that systemd cleans out user
processes on exit now by default.
Might not be exactly the same, the article I remembered talks about that
behavior even killing screen, I think that was KillUserProcess instead of
the shared memory setting, but very similar:
https://www.phoronix.com/scan.php?page=news_item&px=systemd-230-KillUse…
I think maybe the jack FAQ needs a section for headless use, it seems
about twice a year some problem comes up and it takes a few days to work
through all the variations of PAM and systemd caveats needed to get jack
working on an appliance style device.
--
Chris Caudle
2:04 p.m.
New subject: [Jack-Devel] Usage feasibility Q
On 2018-01-31 21:46, Chris Caudle wrote:
<snip />
I think maybe the jack FAQ needs a section for headless use, it seems
about twice a year some problem comes up and it takes a few days to work
through all the variations of PAM and systemd caveats needed to get jack
working on an appliance style device.
That'd be lovely. FWIW, I've been playing with JACK on a Pi as well, and
ran into the dbus issue. Yes, it took quite some digging, some of it in
mailing list archives -- which won't be around forever, to get around it.
--
M.
7:09 p.m.
New subject: [Jack-Devel] Usage feasibility Q
I think maybe the jack FAQ needs a section for
headless use, it seems
about twice a year some problem comes up and it takes a few days to work
through all the variations of PAM and systemd caveats needed to get jack
working on an appliance style device.
Which would be great. I've put together everything I needed to do in order to get it
working: http://forum.audioinjector.net/viewtopic.php?f=5&t=2727&start=30#p5…
That link *should* be publicly accessible.
Regards
/Robert
1:58 p.m.
New subject: [Jack-Devel] Usage feasibility Q
On Wed, January 31, 2018 2:19 pm, David Kastrup wrote:
Never mind. Saw in another answer that systemd's
idea of ending a user
session is at fault.
One thing I don't quite get, according to the logind.conf man page, the
RemoveIPC setting as true will cause the shared resources to be removed
"when the user fully logs out," but does no define how "fully logs
out" is
determined. I would have thought that the user shell still running jackd
would cause "fully logged out" to evaluate to false.
Is there some quirk to how the shell is started from runuser that would
make logind determine the user was not actually logged in? Maybe because
runuser was started from sudo and so did not login with a password?
--
Chris Caudle
2:33 p.m.
New subject: [Jack-Devel] Usage feasibility Q
On Wed, Jan 31, 2018 at 09:19:09PM +0100, David Kastrup wrote:
Never mind. Saw in another answer that systemd's
idea of ending a user
session is at fault. IIRC, systemd even overrides an explicit nohup
anyway (WTF?!?).
This stuff makes me glad I'm still using sysvinit on my
Debian systems.
John
2277
days inactive
2277
days old
jackaudio@lists.linuxaudio.org
16 comments
6 participants
participants (6)
-
Chris Caudle -
David Kastrup -
John Rigg -
Malik Costet -
Ralf Mattes -
Robert Bielik