Re: Too many pythons in poudriere

From: Mark Millard via freebsd-ports <freebsd-ports_at_freebsd.org>
Date: Sun, 11 Jul 2021 01:56:49 UTC
On 2021-Jul-10, at 15:54, bob prohaska <fbsd at www.zefox.net> wrote:

> On Sat, Jul 10, 2021 at 02:17:46PM -0700, Mark Millard wrote:
>> Your latest build did not run out of swap/paging space.
>> It just hit the time limits that you have set, in this
>> case a 172800 second one:
>> 
>> . . .
>> [49:09:17] [ 40% 19569/47953] c++ . . . -o obj/net/third_party/quiche/quiche/quic_server_initiated_spdy_stream.o
>> [49:09:39] =>> Killing timed out build after 172800 seconds
>> [49:09:47] [ 40% 19570/47953] c++ . . .-o obj/net/third_party/quiche/quiche/quic_server_session_base.o
>> . . .
>> 
>> It gave the build a couple of hours before it forced the
>> cleanup activity:
>> 
>> [51:20:04] =>> Cleaning up wrkdir
>> . . .
>> [51:20:06] ===>  Cleaning for chromium-91.0.4472.114_1
>> [51:20:07] [ 44% 21217/47953] python . . . --enable-feature is_linux
>> 
>> which forced a failure to happen from missing files:
>> 
>> [51:20:07] FAILED: gen/ui/events/mojom/event.mojom-module gen/ui/events/mojom/event_constants.mojom-module gen/ui/events/mojom/keyboard_codes.mojom-module gen/ui/events/mojom/scroll_granularity.mojom-module 
>> [51:20:07] python ../../mojo/public/tools/mojom/mojom_parser.py --input-root /wrkdirs/usr/ports/www/chromium/work/chromium-91.0.4472.114 --input-root /wrkdirs/usr/ports/www/chromium/work/chromium-91.0.4472.114/out/Release/gen --output-root /wrkdirs/usr/ports/www/chromium/work/chromium-91.0.4472.114/out/Release/gen --mojom-file-list=__ui_events_mojom_mojom__parser___build_toolchain_linux_clang_arm64__rule..rsp --check-imports /wrkdirs/usr/ports/www/chromium/work/chromium-91.0.4472.114/out/Release/gen/ui/events/mojom/mojom.build_metadata --enable-feature is_posix --enable-feature is_linux
>> [51:20:07] Traceback (most recent call last):
>> [51:20:07]   File "../../mojo/public/tools/mojom/mojom_parser.py", line 456, in <module>
>> [51:20:07]   File "../../mojo/public/tools/mojom/mojom_parser.py", line 448, in Run
>> [51:20:07]   File "../../mojo/public/tools/mojom/mojom_parser.py", line 280, in _ParseMojoms
>> [51:20:07]   File "../../mojo/public/tools/mojom/mojom_parser.py", line 62, in _ResolveRelativeImportPath
>> [51:20:07] ValueError: "mojo/public/mojom/base/time.mojom" does not exist in any of ['/wrkdirs/usr/ports/www/chromium/work/chromium-91.0.4472.114', '/wrkdirs/usr/ports/www/chromium/work/chromium-91.0.4472.114/out/Release/gen']
>> [51:20:07] ninja: build stopped: subcommand failed.
>> [51:20:08] *** Error code 1
>> [51:20:08] 
>> [51:20:08] Stop.
>> [51:20:08] make: stopped in /usr/ports/www/chromium
>> [51:24:17] build of www/chromium | chromium-91.0.4472.114_1 ended at Fri Jul  9 23:39:24 PDT 2021
>> [51:24:17] build time: 51:24:16
>> [51:24:17] !!! build failure encountered !!!
>> 
>> It looks like, in addition to other aspects of getting a
>> coherent and clean environment going, you will need to
>> increase that 172800 (and possibly other timeout settings).
>> 
>> Otherwise there is the swap space use vs. ALLOW_MAKE_JOBS
>> and MAKE_JOBS_NUMBER sorts of tradeoffs for the 1 GiByte RAM
>> RPi3B, even for a coherent/clean build context.
>> 
> 
> Ahh,  I didn't notice that it was a timeout and thought
> something serious triggered the "subcommand failed" message. 

Where:

http://www.zefox.org/~bob/poudriere/data/logs/bulk/main-default/2021-07-07_20h08m41s/build.html

reports:

Phase build/timeout
and:
Log runaway_process

that should indicate that a timeout has occurred and
should be able to be found in the log if I understand
right. It is possible that there is more than a
timeout of interest in the log, even before the
timeout.

As can be seen by the 2 hr delay in the example, the
timeout need not be near the final activity.

> As you've guided me through poudriere I'm starting realize
> it's a textile mill and my goal calls for a needle and thread. 

I'd never call building www/chromium and what it depends on
as something on the scale of "a needle and thread", even on
a machine with lots of resources for the activity --and so
with fewer complications to deal with.

===
Mark Millard
marklmi at yahoo.com
( dsl-only.net went
away in early 2018-Mar)