RGW's init script on RHEL runs the daemon as the "apache" UID, but it's clear that this configuration was not as well-tested as it should have been. During the 1.2.3 release cycle we had multiple bugs crop up due to this configuration: - bug 1184588 - bug 1202784 - others I am probably forgetting about Still outstanding issues unfixed in 1.2.3: - bug 1203926 , problem with log file permissions - bug where the socket directory gets removed on every reboot in RHEL 7 Other aspects to consider: - Debian's init systems have always run RGW as root. There was a brief change upstream, but it was quickly reverted (see http://tracker.ceph.com/issues/9001) - We will have to switch to running RGW as root anyway to support Civetweb in 1.3. (bug 1214518)
https://github.com/ceph/ceph/pull/4638 shipped in Ceph v0.80.10 upstream, so we'll take in in as part of the RHCS 1.2 rebase to Ceph v0.80.11 (bz 1271774)
Ken, what is the list of things that has to be covered as part of this bug verification?
Basically sanity-test that "service ceph-radosgw start" works on RHEL 6 and RHEL 7, both before and after rebooting, etc. `ps -elf | grep radosgw` should show that the radosgw service is running as root, not apache.
Closing all RHCS 1.2 bugs, since we are not planning to do a formal RHCS 1.2.4 release. This particular issue was already fixed in RHCS 1.3, in https://bugzilla.redhat.com/show_bug.cgi?id=1214518