[lttng-dev] babeltrace2 python api related question
In command line, I use “babeltrace2 net://somehost/some-session” to read live
lttng session trace, and I want to use python bindings to do the same, is it
possible?
here is my code:
import bt2
import sys
# Create an empty graph.
graph = bt2.Graph()
# Add a `source.text.dmesg` component.
#
# graph.add_component() returns the created and added component.
#
# Such a component reads Linux kernel ring buffer messages (see
# `dmesg(1)`) from the standard input and creates corresponding event
# messages. See `babeltrace2-source.text.dmesg(7)`.
#
# `my source` is the unique name of this component within `graph`.
# the original source code here in the example is:
# comp_cls = bt2.find_plugin('text').source_component_classes['dmesg']
# src_comp = graph.add_component(comp_cls,'my source')
comp_cls = bt2.find_plugin('ctf').source_component_classes['lttng-live']
src_comp = graph.add_component(comp_cls, 'my source', params = {
'inputs': [sys.argv[1]],
})
# Add a `sink.text.pretty` component.
#
# Such a component pretty-prints event messages on the standard output
# (one message per line). See `babeltrace2-sink.text.pretty(7)`.
#
# The `babeltrace2 convert` CLI command uses a `sink.text.pretty`
# sink component by default.
comp_cls = bt2.find_plugin('text').sink_component_classes['pretty']
sink_comp = graph.add_component(comp_cls, 'my sink')
# Connect the `out` output port of the `source.text.dmesg` component
# to the `in` input port of the `sink.text.pretty` component.
graph.connect_ports(src_comp.output_ports['out'], sink_comp.input_ports['in'])
# Run the trace processing graph.
graph.run()
发件人: Philippe Proulx
发送时间: 2022年8月10日 22:08
收件人: zenan
抄送: lttng-dev@lists.lttng.org
主题: Re: [lttng-dev] babeltrace2 python api related question
On Wed, Aug 10, 2022 at 9:54 AM zenan via lttng-dev
wrote:
>
>
>
> Hi there,
>
> I’m trying to use babeltrace2 python bindings to track live lttng trace, as
> shown in some examples in the documents, I tried to use ctf.lttng-live(
> ctf.fs in the documents) plugin as source, and use text.pretty as sink, but
> it didn’t work out as I expected. So does the api support tracking live
> lttng trace and print them out yet?
>
>
What did you try exactly?
I need command lines and/or some code (keep it minimal) to help you.
Philippe
>
>
>
> Yours sincerely
>
> ___
> lttng-dev mailing list
> lttng-dev@lists.lttng.org
> https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
___
lttng-dev mailing list
lttng-dev@lists.lttng.org
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
Re: [lttng-dev] Change to CPU number detection breaks QNX compilation
On 2022-08-13 19 h 10, Elad Lahav via lttng-dev wrote: Hello, I have just downloaded and built the library for QNX/aarch64 and noticed that the code in compat-smp.h breaks the build. This file appears to be a (very) recent addition, from this commit: --- commit 5cfe81b7ddff9543d451746de9965cac58c67182 Author: Michael Jeanson Date: Wed Jul 27 10:44:00 2022 -0400 fix: sysconf(_SC_NPROCESSORS_CONF) can be less than max cpu id --- I started working around it, but it got a bit ugly. Is the change not a Linuxism? Does it make sense on other systems? If QNX is the exception then I can try to come up with a patch, but if not perhaps the entire code should be wrapped by #ifdef __linux__? --Elad Hi, We'll need more details on the build failure you are seeing, there is already an '#ifdef __linux__' in 'src/compat-smp.h'. Can you send the relevant build errors? Michael ___ lttng-dev mailing list lttng-dev@lists.lttng.org https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
Re: [lttng-dev] Change to CPU number detection breaks QNX compilation
On 2022-08-15 10 h 04, Elad Lahav wrote: The problem is the use of d_type as a member of the dirent structure. POSIX does not define such a member. I can provide the necessary code for getting the file type in a readdir() loop on QNX, but it seems unnecessary given that there is no /sys file system with CPU information. The method for getting information on processors is different, and I suspect that is true for *BSDs and other non-Linux OSs. --Elad Can you try this patch [1] to see if it fixes the build on QNX? Can you also run 'make check' in 'tests/unit', the linux specific tests should be skipped. Michael [1] https://review.lttng.org/c/userspace-rcu/+/8664 ___ lttng-dev mailing list lttng-dev@lists.lttng.org https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
Re: [lttng-dev] Change to CPU number detection breaks QNX compilation
That fixes the build, thanks. I see some errors coming out of "make check", though: ERROR: test_get_cpu_mask_from_sysfs - missing test plan ERROR: test_get_cpu_mask_from_sysfs - exited with status 1 --Elad On Mon, 15 Aug 2022 at 11:55, Michael Jeanson wrote: > > On 2022-08-15 10 h 04, Elad Lahav wrote: > > The problem is the use of d_type as a member of the dirent structure. > > POSIX does not define such a member. > > I can provide the necessary code for getting the file type in a > > readdir() loop on QNX, but it seems unnecessary given that there is no > > /sys file system with CPU information. The method for getting > > information on processors is different, and I suspect that is true for > > *BSDs and other non-Linux OSs. > > > > --Elad > > Can you try this patch [1] to see if it fixes the build on QNX? Can you > also run 'make check' in 'tests/unit', the linux specific tests should > be skipped. > > Michael > > [1] https://review.lttng.org/c/userspace-rcu/+/8664 ___ lttng-dev mailing list lttng-dev@lists.lttng.org https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
Re: [lttng-dev] Change to CPU number detection breaks QNX compilation
On 2022-08-15 12 h 56, Elad Lahav wrote: That fixes the build, thanks. I see some errors coming out of "make check", though: ERROR: test_get_cpu_mask_from_sysfs - missing test plan ERROR: test_get_cpu_mask_from_sysfs - exited with status 1 --Elad I tested on Freebsd (as a non-linux target) and I get the expected : SKIP: test_get_cpu_mask_from_sysfs - Skip Linux specific tests. I wonder what might differ here, it might have to do with the version of bash on QNX, I'm not familiar with this platform. Michael ___ lttng-dev mailing list lttng-dev@lists.lttng.org https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
Re: [lttng-dev] Change to CPU number detection breaks QNX compilation
/bin/sh on QNX is (by default) ksh, not bash, so perhaps that's the issue. Shouldn't block you from submitting that change. +1 from me, if that matters ;-) --Elad On Mon, 15 Aug 2022 at 13:39, Michael Jeanson wrote: > > On 2022-08-15 12 h 56, Elad Lahav wrote: > > That fixes the build, thanks. > > > > I see some errors coming out of "make check", though: > > > > ERROR: test_get_cpu_mask_from_sysfs - missing test plan > > ERROR: test_get_cpu_mask_from_sysfs - exited with status 1 > > > > --Elad > > I tested on Freebsd (as a non-linux target) and I get the expected : > > SKIP: test_get_cpu_mask_from_sysfs - Skip Linux specific tests. > > I wonder what might differ here, it might have to do with the version of > bash on QNX, I'm not familiar with this platform. > > Michael ___ lttng-dev mailing list lttng-dev@lists.lttng.org https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
Re: [lttng-dev] babeltrace2 python api related question
On Sun, Aug 14, 2022 at 4:48 AM zenan wrote:
>
> In command line, I use “babeltrace2 net://somehost/some-session” to read live
> lttng session trace, and I want to use python bindings to do the same, is it
> possible?
>
> here is my code:
Almost there.
Some `src.ctf.lttng-live` component and message iterator methods may
return a "try again" status code (I don't remember exactly under which
circumstances). This propagates to bt_graph_run() [1] (in C) eventually
which itself returns `BT_GRAPH_RUN_STATUS_AGAIN`. The Python
bt2.Graph.run() method translates this status code to throwing
`bt2.TryAgain`.
Just catch those `bt2.TryAgain`, wait a bit, and call `graph.run()`
again:
# Run the trace processing graph.
while True:
try:
graph.run()
except bt2.TryAgain:
time.sleep(.1)
Works for me.
If you ever need to access the messages themselves instead of forwarding
them to a `sink.text.pretty` component, then use a
`bt2.TraceCollectionMessageIterator` instance [2].
Hope it helps,
Phil
[1]:
https://babeltrace.org/docs/v2.0/libbabeltrace2/group__api-graph.html#gad2e1c1ab20d1400af1b552e70b3d567c
[2]:
https://babeltrace.org/docs/v2.0/python/bt2/examples.html#iterate-trace-events
>
> import bt2
> import sys
>
> # Create an empty graph.
> graph = bt2.Graph()
>
> # Add a `source.text.dmesg` component.
> #
> # graph.add_component() returns the created and added component.
> #
> # Such a component reads Linux kernel ring buffer messages (see
> # `dmesg(1)`) from the standard input and creates corresponding event
> # messages. See `babeltrace2-source.text.dmesg(7)`.
> #
> # `my source` is the unique name of this component within `graph`.
>
> # the original source code here in the example is:
> # comp_cls = bt2.find_plugin('text').source_component_classes['dmesg']
> # src_comp = graph.add_component(comp_cls,'my source')
> comp_cls = bt2.find_plugin('ctf').source_component_classes['lttng-live']
> src_comp = graph.add_component(comp_cls, 'my source', params = {
> 'inputs': [sys.argv[1]],
> })
>
> # Add a `sink.text.pretty` component.
> #
> # Such a component pretty-prints event messages on the standard output
> # (one message per line). See `babeltrace2-sink.text.pretty(7)`.
> #
> # The `babeltrace2 convert` CLI command uses a `sink.text.pretty`
> # sink component by default.
> comp_cls = bt2.find_plugin('text').sink_component_classes['pretty']
> sink_comp = graph.add_component(comp_cls, 'my sink')
>
> # Connect the `out` output port of the `source.text.dmesg` component
> # to the `in` input port of the `sink.text.pretty` component.
> graph.connect_ports(src_comp.output_ports['out'], sink_comp.input_ports['in'])
>
> # Run the trace processing graph.
> graph.run()
>
>
>
>
>
> 发件人: Philippe Proulx
> 发送时间: 2022年8月10日 22:08
> 收件人: zenan
> 抄送: lttng-dev@lists.lttng.org
> 主题: Re: [lttng-dev] babeltrace2 python api related question
>
>
>
> On Wed, Aug 10, 2022 at 9:54 AM zenan via lttng-dev
>
> wrote:
>
> >
>
> >
>
> >
>
> > Hi there,
>
> >
>
> > I’m trying to use babeltrace2 python bindings to track live lttng trace, as
> > shown in some examples in the documents, I tried to use ctf.lttng-live(
> > ctf.fs in the documents) plugin as source, and use text.pretty as sink, but
> > it didn’t work out as I expected. So does the api support tracking live
> > lttng trace and print them out yet?
>
> >
>
> >
>
>
>
> What did you try exactly?
>
>
>
> I need command lines and/or some code (keep it minimal) to help you.
>
>
>
> Philippe
>
>
>
> >
>
> >
>
> >
>
> > Yours sincerely
>
> >
>
> > ___
>
> > lttng-dev mailing list
>
> > lttng-dev@lists.lttng.org
>
> > https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
>
>
___
lttng-dev mailing list
lttng-dev@lists.lttng.org
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
