----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/54212/#review157458 -----------------------------------------------------------
src/slave/containerizer/mesos/provisioner/backends/overlay.cpp (line 167) <https://reviews.apache.org/r/54212/#comment228051> Do we want to consider maintain a similar hierarchy as `scratch`? It seems asymmetrical if symlinks and scratch is at different file system level. Maybe we should consider make this `path::join(backendDir, "links", rootfsId)`? - Zhitao Li On Nov. 30, 2016, 6:54 p.m., Gilbert Song wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/54212/ > ----------------------------------------------------------- > > (Updated Nov. 30, 2016, 6:54 p.m.) > > > Review request for mesos, Artem Harutyunyan, Jie Yu, Qian Zhang, and Zhitao > Li. > > > Bugs: MESOS-6653 > https://issues.apache.org/jira/browse/MESOS-6653 > > > Repository: mesos > > > Description > ------- > > Since the fix of MESOS-6000, symlinks are used in overlayfs > backend to shorten the arguments when mounting the rootfs. > E.g., '.../backends/overlay/links' is the symlink created > for a provisioned image. It becomes problematic if a > container image is specified while some image volumes are > specified for the same container. An unique symlink is > needed for each image to be provisioned. > > > Diffs > ----- > > src/slave/containerizer/mesos/provisioner/backends/overlay.cpp > b71a31323aef376c9a28e1d52322a1802fb212ad > src/tests/containerizer/provisioner_backend_tests.cpp > 727bf645dd9337a94f098ab0a816b7e0e0651083 > > Diff: https://reviews.apache.org/r/54212/diff/ > > > Testing > ------- > > make check > > Verified that the unit test `ROOT_ImageInVolumeWithRootFilesystem` passed. > > > Thanks, > > Gilbert Song > >
