-
Note to everyone interested in helping in testing update candidates (which is very important for the confidence we have in our updates), you can subscribe to this topic and make sure your user settings in this forum allow sending notification e-mails to you (it does not by default).
Update candidate:
xcp-emu-manager
- target: XCP-ng 7.5 and XCP-ng 7.6
- reasons for the update: fix a VM migration issue that occurs mostly with loaded VMs
- packages:
- 7.6:
xcp-emu-manager-0.0.6-1.x86_64.rpm
- 7.5:
xcp-emu-manager-0.0.3-1.1.x86_64.rpm
- 7.6:
- description of the changes:
- Many users reported seemingly random migration failures. Community testing alllowed to find that it was consistently reproducible by forcing high resource usage such as with this command:
stress --cpu 1
- The problem was caused by the fact that
xcp-emu-manager
, our replacement of the proprietaryemu-manager
component from Citrix that handles suspend, resume and migrate operations, did not know how to handle progress messages that arrived when not expected anymore, near the end of the migration process. The fix consists in ignoring those messages when the next step of the process has already started. - Fixes https://github.com/xcp-ng/xcp/issues/72
- Many users reported seemingly random migration failures. Community testing alllowed to find that it was consistently reproducible by forcing high resource usage such as with this command:
- install:
yum install --enablerepo='xcp-ng-updates_testing' xcp-emu-manager
- testing tips: migrate VMs, especially loaded ones. Or artificially stress linux VMs with the
stress
command then migrate them. - reboot: no reboot required, no toolstack restart.
For this specific update, please report your results there: https://xcp-ng.org/forum/topic/522/unable-to-migrate-live-vms-after-upgrading-from-xcp-ng-7-4-to-7-5
-
The update for 7.6 has been made available as a regular update, as
xcp-emu-manager-0.0.6-2.x86_64.rpm
. It should prevent most migration crashes, but can still get stuck when the VM is heavily loaded. Instead of failing directly, it will now try to wait for a low enough activity to resume the process. We're still working on it to bring a better, complete solution. -
Update candidate:
xen
- target: XCP-ng 7.5 and XCP-ng 7.6
- reasons for the update: security update
- packages:
- 7.6:
xen-debuginfo-4.7.6-7.1.1.xcp.x86_64.rpm xen-devel-4.7.6-7.1.1.xcp.x86_64.rpm xen-dom0-libs-4.7.6-7.1.1.xcp.x86_64.rpm xen-dom0-libs-devel-4.7.6-7.1.1.xcp.x86_64.rpm xen-dom0-tools-4.7.6-7.1.1.xcp.x86_64.rpm xen-hypervisor-4.7.6-7.1.1.xcp.x86_64.rpm xen-hypervisor-debuginfo-4.7.6-7.1.1.xcp.x86_64.rpm xen-installer-files-4.7.6-7.1.1.xcp.x86_64.rpm xen-libs-4.7.6-7.1.1.xcp.x86_64.rpm xen-libs-devel-4.7.6-7.1.1.xcp.x86_64.rpm xen-ocaml-devel-4.7.6-7.1.1.xcp.x86_64.rpm xen-ocaml-libs-4.7.6-7.1.1.xcp.x86_64.rpm xen-tools-4.7.6-7.1.1.xcp.x86_64.rpm
- 7.5: same packages in version 4.7.5-5.7.1.xcp
- description of the changes:
- Security fixes from the Xen project
- https://support.citrix.com/article/CTX239432
- install:
yum install --enablerepo='xcp-ng-updates_testing' xen-dom0-libs xen-hypervisor xen-tools xen-dom0-tools xen-libs
or if you don't mind installing other update candidates at the same time:yum update --enablerepo='xcp-ng-updates_testing'
- testing tips: basic functions of the hypervisor
- reboot: yes
-
An update to XCP-ng 7.6 has been pushed:
xcp-emu-manager-0.0.9-1
, that fixes all known migration issues (well, except this bug from XenServer: https://github.com/xcp-ng/xcp/issues/111). Update for XCP-ng 7.5 is being prepared. -
Updates to XCP-ng 7.6 have been pushed today:
xapi-*
andxcp-featured
: add ZSTD compression support for faster export and backupsopenvswitch
: fix wrong log rotation configuration (https://github.com/xcp-ng/xcp/issues/100)
Reminder: how to update.
-
New update candidate for XCP-ng 7.6. It adds support for NFS 4.1, that has been contributed by a member of the community and already merged upstream in the XAPI project!
Install it:
yum update sm sm-rawhba --enablerepo='xcp-ng-updates_testing'
Revert:
yum downgrade sm sm-rawhba
Thanks in advance for your feeback!
-
I feel alone in this thread
-
U are never ALONE
-
@stormi I am watching you.
-
We are the watchmens
-
I'm here. Alive n' kickin'
-
So did anyone install the update candidate for
sm
, check that they see no obvious regression and ideally test NFS 4.1 support? -
@stormi
I have an netapp storage. it support nfs 4.1. how can I add a nfs sr with nfs 4.1? I want test it. -
@stormi I did not test it (yet) .. currently copying loads of vm data around at work... but if all goes well I have a free storage next week that will become our new nfs4.1 server for work ... stay tuned
-
@rizaemet-0 here is a XCP-ng Center with NFS 4.1 support: https://github.com/xcp-ng/xenadmin/releases/tag/v7.6.3.21
Just add your NFS SR via the normal assistant, the NFS 4.1 option is right there.
-
I create a nfs sr with nfs 4.1 with xcp-ng center v7.6.3.21. success.
live storage migrate a vm to this sr. success.
live vm migrate to another host. success.
backup vm with xo. success.
I tested copy-paste file in itself and nfs4, nfs4.1, fc between each other. nfs 4.1 speed is like nfs4 speed for my environment.
heavy i/o for 24 hours. no any performance or another issue. -
Do you have better perfs than 4.0 by the way?
-
@olivierlambert Similar performance for my environment. I did not any specific test. Just copy-paste files.
By the way, is multipath connection (or pnfs) possible xcp-ng between storage with nfs4.1? -
It should be possible if you SR is supporting pNFS (because IIRC, it's included in 4.1 protocol).
However, pNFS doesn't remove the usual SPOF (metadata NFS server)
-
pNFS may not be available in XCP-NG's kernel. I see option
CONFIG_NFSD_PNFS
unset where as most of the distros haveCONFIG_NFSD_PNFS=y