Bug 1127457
Summary: | Setting security.* xattrs fails | |||
---|---|---|---|---|
Product: | [Community] GlusterFS | Reporter: | Diego Woitasen <diego> | |
Component: | fuse | Assignee: | bugs <bugs> | |
Status: | CLOSED EOL | QA Contact: | ||
Severity: | high | Docs Contact: | ||
Priority: | unspecified | |||
Version: | 3.5.2 | CC: | bugs, diego, jbyers, lotz, madam, ndevos, pille+redhat+bugzilla, prasun.gera | |
Target Milestone: | --- | Keywords: | EasyFix, Triaged | |
Target Release: | --- | |||
Hardware: | x86_64 | |||
OS: | Linux | |||
Whiteboard: | ||||
Fixed In Version: | Doc Type: | Bug Fix | ||
Doc Text: | Story Points: | --- | ||
Clone Of: | ||||
: | 1283103 (view as bug list) | Environment: | ||
Last Closed: | 2016-06-17 15:57:15 UTC | Type: | Bug | |
Regression: | --- | Mount Type: | --- | |
Documentation: | --- | CRM: | ||
Verified Versions: | Category: | --- | ||
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | ||
Cloudforms Team: | --- | Target Upstream Version: | ||
Embargoed: | ||||
Bug Depends On: | 1283103 | |||
Bug Blocks: |
Description
Diego Woitasen
2014-08-06 22:03:39 UTC
More information, my fstab line: storage01:/gv0 /srv/data glusterfs defaults,acl 0 0 Adding user_xattr doesn't work, I can't mount the volume with that option. BTW, I can set xattrs, the problem is with security prefixed xattrs. From xlators/mount/fuse/src/fuse-bridge.c: 3136 if (!priv->selinux) { 3137 if (strncmp (name, "security.", 9) == 0) { 3138 send_fuse_err (this, finh, EOPNOTSUPP); 3139 GF_FREE (finh); 3140 return; 3141 } 3142 } Checking a little further, it seems that you can mount with the 'selinux' option. In /etc/fstab this would then look like this: storage01:/gv0 /srv/data glusterfs defaults,acl,selinux 0 0 I understand that Ubuntu does not use selinux, but that does not matter for using this mount option. Could you test and report if that works for you? Proposed fix: We should not disallow setting security.* xattrs when selinux is not in use, selinux is not the only component that uses security.* xattrs. I think that instead of filtering out all security.* xattrs, only security.selinux should be denied. Adding the selinux mount options it works. Thanks for your help. i'm currently ingesting several TB into our glusterfs-cluster and my logfile grows multiple GB/day because of (three times per file): [2015-01-23 10:31:04.835822] E [marker.c:2542:marker_removexattr_cbk] 0-storage_www-marker: No data available occurred while creating symlinks [2015-01-23 10:31:04.835850] I [server-rpc-fops.c:693:server_removexattr_cbk] 0-storage_www-server: 88351: REMOVEXATTR /file1234 (85c82646-7c31-4773-ae5f-fa7661452f67) of key security.ima ==> (No data available) mounting using -o selinux doesn't help. i'm on ubuntu 14.04, too. glusterfs is v3.6.1. The missing (No data available) security.ima extended attribute might be a kernel issue. I think this patch (or some follow up) fixes it: https://lkml.org/lkml/2014/7/11/259 It would be interesting to know if Ubuntu 14.04 has these patches applied. Alternatively, you can probably disable the IMA function in Ubuntu to prevent these messages. Could you let us know if that helps? i couldn't figure out, whether these patches are applied to the stock ubuntu kernel, but found nothing in the changelog. but i installed a 14.10 kernel (3.16), which should have these commits already in. unfortunately the behavior didn't change. i didn't try to disable IMA yet. Using the -o selunux mount option does not seem to be a good work-around, at least not when selinux is not in use, and Gluster replica volumes are in use. In GlusterFS 3.6.1, copying a directory tree takes more than 600% longer when the FUSE mount has the selinux option than it does without it. This seems to be due to all of the: fuse-bridge.c:3335:fuse_xattr_cbk ... GETXATTR(security.capability) ... (No data available) failures. My logs are filled with these messages too on RHS 3.0 on RHEL 6.7. Also, replacing a brick and heal is taking exceptionally long. Don't know if they are related. On replacing a brick in distributed replicated volume, it had managed to move only 100G in 24 hours. This bug is getting closed because the 3.5 is marked End-Of-Life. There will be no further updates to this version. Please open a new bug against a version that still receives bugfixes if you are still facing this issue in a more current release. |