summaryrefslogtreecommitdiffstats
path: root/bitbake/lib/bb/runqueue.py
diff options
context:
space:
mode:
authorMark Asselstine <mark.asselstine@windriver.com>2023-12-28 16:01:17 -0500
committerRichard Purdie <richard.purdie@linuxfoundation.org>2024-01-10 14:02:38 +0000
commitbc22d82c2f0c16ee8ad2edeb3bcdfb31db0afa94 (patch)
tree5b2ec6128e29913c02e4434e1756afc5b4b92b07 /bitbake/lib/bb/runqueue.py
parentc665a2c93337fc372207af604c21bdea5babb4e5 (diff)
downloadpoky-bc22d82c2f0c16ee8ad2edeb3bcdfb31db0afa94.tar.gz
bitbake: server/process: catch and expand multiprocessing connection exceptions
Doing builds on systems with limited resources, or with high demand package builds such as chromium it isn't uncommon for the OOM Killer to be triggered and for bitbake-server to be selected as the process to be killed. When the bitbake-server does terminate unexpectedly due to the OOM Killer or otherwise, this currently results in a generic python traceback with little indication as to what has failed. Here we trap and raise the exceptions while extending the exception text in runCommand() to make it clear that this is most likely caused by the bitbake-server unexpectedly terminating. Callers of runCommand() should be updated to properly handle the BrokenPipeError and EOFError exceptions to avoid printing a python traceback, but even if they don't, the added text in the exceptions should provide some hints as to what might have caused the failure. (Bitbake rev: 5ff62b802f79acc86bbd6a99484f08501ff5dc2d) Signed-off-by: Mark Asselstine <mark.asselstine@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Diffstat (limited to 'bitbake/lib/bb/runqueue.py')
0 files changed, 0 insertions, 0 deletions