Description of problem: In 3.2 VDSM talked to ksm via ksmtuned. 3.3 VDSM uses MOM to talk to ksm, but upgrading VDSM from 3.2 to 3.3 doesn't disable ksmtuned. That means that VDSM may send conflicting instructions to KSM as both ksmtuned and MOM are active. In systemd we marked VDSM as conflicting with ksmtuned, so something similar needs to happen with systemV. Additionally, there is a nontrivial amount of code in VDSM that can now be removed with MOM now managing ksm. Version-Release number of selected component (if applicable): oVirt Engine Version: 3.3.0-0.3.beta1.fc19 VDSM 4.12.0-0.1.rc3
*** Bug 987921 has been marked as a duplicate of this bug. ***
Running the following commands on host upgraded from 3.2 to 3.3: # service --status-all | grep ksm ksm is not running ksmtuned (pid 8354) is running... # chkconfig | grep ksm ksm 0:off 1:off 2:off 3:on 4:on 5:on 6:off ksmtuned 0:off 1:off 2:off 3:on 4:on 5:on 6:off it seems that the ksmtuned service is not disabled, but ignored. Is this enough ? shouldn't we turned it off + kill it ?
ksmtuned is stopped when vdsm starts (http://gerrit.ovirt.org/#/c/17537/). I do not think that "simple" installation of a package should change the global system settings.
(In reply to Martin Sivák from comment #3) > ksmtuned is stopped when vdsm starts (http://gerrit.ovirt.org/#/c/17537/). I # service --status-all | egrep -i "ksm|vdsm" ksm is not running ksmtuned (pid 8354) is running... # service vdsmd status VDS daemon server is running > do not think that "simple" installation of a package should change the > global system settings.
That is weird. Our vdsm's initscript should stop ksmtuned. But it should not touch ksm at all. Can you: 1) stop vdsm 2) start ksm 3) start ksmtuned 4) start vdsm and check the status after each step? What you describe is quite unusual. ksm should be on at all times. And btw, was vdsm restarted after the upgrade?
(In reply to Martin Sivák from comment #5) > That is weird. Our vdsm's initscript should stop ksmtuned. But it should not > touch ksm at all. > > Can you: > 1) stop vdsm # service vdsmd status VDS daemon is not running # service --status-all | egrep 'ksm|ksmtuned' ksm is not running ksmtuned is stopped > 2) start ksm # service vdsmd status VDS daemon is not running # service --status-all | egrep 'ksm|ksmtuned' ksm is running ksmtuned is stopped > 3) start ksmtuned # service vdsmd status VDS daemon is not running # service --status-all | egrep 'ksm|ksmtuned' ksm is running ksmtuned (pid 14894) is running... > 4) start vdsm # service vdsmd status VDS daemon server is running # service --status-all | egrep 'ksm|ksmtuned' ksm is running ksmtuned is stopped > > and check the status after each step? What you describe is quite unusual. > ksm should be on at all times. > > And btw, was vdsm restarted after the upgrade? I'm not sure on what machine I did the verification, but it looks like it did: rose02 ------ 2013-09-04 12:24:40: I am the actual vdsm 4.12.0-105.git0da1561.el6ev 2013-08-28 14:26:07: I am the actual vdsm 4.10.2-24.1.el6ev 2013-08-04 15:10:26: I am the actual vdsm 4.10.2-23.0.el6ev rose03 ------ 2013-09-04 12:26:28: I am the actual vdsm 4.12.0-105.git0da1561.el6ev 2013-08-28 14:25:46: I am the actual vdsm 4.10.2-24.1.el6ev 2013-08-04 14:58:59: I am the actual vdsm 4.10.2-23.0.el6ev
Verified on is13
bulk closing, assuming verified bugs are in 3.3.