I exhausted my virtual memory when trying to install SyntaxNet from this Dockerfile using the Docker Toolbox. I received this message when compiling the Dockerfile:
ERROR: /root/.cache/bazel/_bazel_root/5b21cea144c0077ae150bf0330ff61a0/external/org_tensorflow/tensorflow/core/kernels/BUILD:1921:1: C++ compilation of rule '@org_tensorflow//tensorflow/core/kernels:svd_op' failed: gcc failed: error executing command /usr/bin/gcc -U_FORTIFY_SOURCE '-D_FORTIFY_SOURCE=1' -fstack-protector -Wall -Wl,-z,-relro,-z,now -B/usr/bin -B/usr/bin -Wunused-but-set-parameter -Wno-free-nonheap-object -fno-canonical-system-headers ... (remaining 115 argument(s) skipped): com.google.devtools.build.lib.shell.BadExitStatusException: Process exited with status 1. virtual memory exhausted: Cannot allocate memory ____Building complete. ____Elapsed time: 8548.364s, Critical Path: 8051.91s
I have a feeling this could be resolved by changing Bazel's default jobs limit with (for example) --jobs=1, however I'm not sure where I would put that in the Dockerfile.
There are two possibilities: You could either modify the Dockerfile so that it creates a ~/.bazelrc that contains the following text:
build --jobs=1
Note that this works, even though the Dockerfile runs bazel test (as opposed to bazel build), because build flags in the .bazelrc also apply to Bazel's test command.
The other possibility would be to modify the RUN command in the Dockerfile to include the --jobs=1 parameter, e.g. RUN [...] && bazel test --jobs=1 --genrule_strategy=standalone [...].
Bazel should then spawn not more than a single child process during the build. You can verify this by running "ps axuf" on your host and looking at the process tree of your container. If you modified the RUN cmd, you should also see the --jobs=1 parameter on Bazel's command-line.