Todd Lipcon has posted comments on this change. Change subject: KUDU-1508: enforce block limit on lbm containers ......................................................................
Patch Set 2: (3 comments) http://gerrit.cloudera.org:8080/#/c/5403/2/src/kudu/fs/data_dirs.cc File src/kudu/fs/data_dirs.cc: Line 270: // 1. el6.9: https://bugzilla.redhat.com/show_bug.cgi?id=1351798. sounds like kernel-2.6.32-674.el6 will have the fix for this stream Line 357: RETURN_NOT_OK(FindMountForPath(env_, mounts, p, &found_mount)); instead of parsing the mtab and doing this stuff, why not just use statfs() on the container file and check f_type == EXT4_SUPER_MAGIC? It's slightly less precise since ext3 and ext2 share the same magic number, but i dont think that really matters, since ext3 and ext2 wouldn't support hole punching anyway I think that could simplify the patch a lot: you could just use fstatfs() on the container file when you first open it, and it'll give you both the block size as well as the FS type, so you can set the limit right there and not have to plumb it around through Data Dirs http://gerrit.cloudera.org:8080/#/c/5403/2/src/kudu/util/env_posix.cc File src/kudu/util/env_posix.cc: Line 1277: FILE* f = setmntent(MOUNTED, "r"); apparently this is deprecated, and should use _PATH_MOUNTED instead (https://www.gnu.org/software/libc/manual/html_node/Mount-Information.html) -- To view, visit http://gerrit.cloudera.org:8080/5403 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I63576d2bf2cc61b2deb70f7e166f08e0d4c66543 Gerrit-PatchSet: 2 Gerrit-Project: kudu Gerrit-Branch: master Gerrit-Owner: Adar Dembo <a...@cloudera.com> Gerrit-Reviewer: Dan Burkert <danburk...@apache.org> Gerrit-Reviewer: Kudu Jenkins Gerrit-Reviewer: Tidy Bot Gerrit-Reviewer: Todd Lipcon <t...@apache.org> Gerrit-HasComments: Yes
