Bug 1478122

Summary: [RFE] Move qemu sockets under a tmpfs mount point may be /run instead of /var/lib/libvirt/qemu/channels/
Product: Red Hat Enterprise Virtualization Manager Reporter: Koutuk Shukla <kshukla>
Component: ovirt-engineAssignee: Milan Zamazal <mzamazal>
Status: CLOSED DEFERRED QA Contact: meital avital <mavital>
Severity: medium Docs Contact:
Priority: medium    
Version: 4.1.1CC: branpise, gwatson, kshukla, lsurette, mavital, mtessun, nobody, rbarry, Rhev-m-bugs, srevivo
Target Milestone: ---Keywords: FutureFeature
Target Release: ---Flags: lsvaty: testing_plan_complete-
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Enhancement
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2020-04-01 14:47:18 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: Virt RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Koutuk Shukla 2017-08-03 15:47:04 UTC
1. Proposed title of this feature request  
      -- Move qemu sockets under a tmpfs mount point may be /run instead of /var/lib/libvirt/qemu/channels/

      
2. What is the nature and description of the request?  
--  Is there a possibility to move the qemu sockets from /var/lib/libvirt/qemu/channels/ to a tmpfs mount point may be /run instead. ??

3. Why does the customer need this? (List the business requirements here)  
-- To increase the robustness of the RHV solution and making recovery from similar situations painless.  In this case an I/O mishap caused /var on the nodes to become read-only. If the control sockets have been under /run (a tmpfs mounted fs) they would have been still writable 

4.  What's the advantage of using a different directory ?
-- Named sockets have no writable structure on disk, but their write ability is subject to the write ability of the filesystem where their name is stored. In this case, the socket is used by vdsmd to communicate and control the qemu-kvm processes that implement the VMs in the RHEV system and they don't need to survive reboots of the nodes.   Denying write ability to them impedes any kind of management, even if the VMs are perfectly healthy.
In this case an I/O mishap caused /var on the nodes to become read-only. If the control sockets have been under /run (a tmpfs mounted fs) they would have been still writable (and I'm not aware of any incident that could make /run read only) and we would have been able to migrate the VMs away from the problematic nodes and then reboot them, suffering no downtime at all.

5 How would you like to achieve this? ( Please List the functional requirements here)
-- We see the path '/var/lib/libvirt/qemu/channels/' is created by vdsm/.    I am not sure which configuration file needs to be edited to change the below path is it vdsm.conf or libvirtd.conf.

/usr/lib/python2.7/site-packages/vdsm/constants.py:P_LIBVIRT_VMCHANNELS = '/var/lib/libvirt/qemu/channels/'

    
6. For each functional requirement listed in the previous question, can you test to confirm the requirement is successfully implemented.
 -- Yes

7. Is there already an existing RFE upstream or in Red Hat Bugzilla?  
-- No    

8. Does the customer have any specific time line dependencies and which release would they like to target (i.e. RHEL5, RHEL6)?  
-- No

9. List any affected packages or components.  
 -- NA

Comment 11 Ryan Barry 2019-01-21 14:53:44 UTC
Re-targeting to 4.3.1 since it is missing a patch, an acked blocker flag, or both

Comment 14 Michal Skrivanek 2020-04-01 14:47:18 UTC
ok, closing. Please reopen if still relevant/you want to work on it.

Comment 15 Michal Skrivanek 2020-04-01 14:50:56 UTC
ok, closing. Please reopen if still relevant/you want to work on it.

Comment 18 Red Hat Bugzilla 2024-01-06 04:25:19 UTC
The needinfo request[s] on this closed bug have been removed as they have been unresolved for 120 days