Archive/XCP 1.6 changes compare to 1.1: Difference between revisions

From Xen
Jump to navigationJump to search
 
(10 intermediate revisions by 2 users not shown)
Line 1: Line 1:
::''This is unofficial changelog, describing minor internal changes, which can be interested for ISP or XCP power users''
{{Info|This is a community changelog, describing minor internal changes, which can be interested for ISPs or XCP power users. It typically covers information that were not covered in the product release notes. Feel free to add. We will regularly poll it and migrate items to [http://www.xen.org/download/xcp/releasenotes_1.6.0.html XCP 1.6 release notes].}}

== xapi/core ==
* squeezed service is no longer required for xapi start of VM start. Memory manupulations are require squeezed daemon to be ran.
* guest tools of VMs is no longer required for vm migration/grace shutdown.
* all VM metrics about last shutdown time/reason/initiator (in other-config) is removed. If pool is upgraded or older vm imported, previous data in other-config will stay intact, but will not updates any more (reason: operations moved from xapi to xenops, and those metrics was skipped in process).

== host management ==
* new sequence of host shutdown operation: host-evacuate called automatically, if not, gracefull shutdown sent, if not, forced shutdown. Means: VM survive accidental reboots, but reboot sequence for host is much slower.


==Console ==
==Console ==
Line 7: Line 15:
== OVS ==
== OVS ==
* OVS 1.4.2, main changes:
* OVS 1.4.2, main changes:
** no memory leak on large flow amount
** no memory leak on heavy load with many flows
** normal ipv6 support for ovs-ofctl
** normal ipv6 support for ovs-ofctl
* new service /usr/bin/python /usr/share/openvswitch/scripts/ovs-xapi-sync
* new service /usr/bin/python /usr/share/openvswitch/scripts/ovs-xapi-sync
Line 21: Line 29:
* new log with xen log ring buffer content: /var/log/xen-dmesg
* new log with xen log ring buffer content: /var/log/xen-dmesg
==Xen behavior==
==Xen behavior==
* XCP now disable C-states on CPU with errors in powersaving routines. It show in xen log:
* XCP now disable C-states on CPU with errors in powersaving routines. It appears in xen log:
(XEN) Disabling C-states C3 and C6 on Nehalem Processors due to errata
(XEN) Disabling C-states C3 and C6 on Nehalem Processors due to errata
== VHD/VDI ==
* snapwatchd daemon removed. But VHD tree is coalescing (still not known all cases - at least at the xe sr-scan).
* 'Base copy' now named after the original disk name. No longer 'base copy' in name-label
* vhd coalescing now never remove last base copy (even there is no snapshots or any other reason for splitting). To reduce space usage create new snapshot, delete it, exec xe sr-scan and wait until 'old' and 'new' base copy (disk content prior snapshot creation) will be merged. Disk still gonna continue to grow (base copy is always read only).

= Release Notes Errata =
{{Info|If you find any errors in the [http://www.xen.org/download/xcp/releasenotes_1.6.0.html XCP 1.6 release notes], please let us know here and we will fix these in the static html pages.}}



[[Category:XCP]]
[[Category:XCP]]
[[Category:Release Notes]]

Latest revision as of 21:00, 25 June 2014

Icon Info.png This is a community changelog, describing minor internal changes, which can be interested for ISPs or XCP power users. It typically covers information that were not covered in the product release notes. Feel free to add. We will regularly poll it and migrate items to XCP 1.6 release notes.


xapi/core

  • squeezed service is no longer required for xapi start of VM start. Memory manupulations are require squeezed daemon to be ran.
  • guest tools of VMs is no longer required for vm migration/grace shutdown.
  • all VM metrics about last shutdown time/reason/initiator (in other-config) is removed. If pool is upgraded or older vm imported, previous data in other-config will stay intact, but will not updates any more (reason: operations moved from xapi to xenops, and those metrics was skipped in process).

host management

  • new sequence of host shutdown operation: host-evacuate called automatically, if not, gracefull shutdown sent, if not, forced shutdown. Means: VM survive accidental reboots, but reboot sequence for host is much slower.

Console

  • new command: xe console vm=....
  • vm: other-config:disable_pv_vnc is no longer supported
  • vncterm can provide telnet-like access to VM consoles (PV only?), called 'tc'. Port can be found in xenstore /local/domain/ID/console/tc-port.

OVS

  • OVS 1.4.2, main changes:
    • no memory leak on heavy load with many flows
    • normal ipv6 support for ovs-ofctl
  • new service /usr/bin/python /usr/share/openvswitch/scripts/ovs-xapi-sync

IP/MAC antispoofing

  • new values in vif:
               locking-mode ( RW): locked
               ipv4-allowed (SRW): X.X.X.X
               ipv6-allowed (SRW): YY:YY:YY::YY:YY
  • new vif initialization script: /opt/xensource/libexec/setup-vif-rules

Logging

  • new log with xen log ring buffer content: /var/log/xen-dmesg

Xen behavior

  • XCP now disable C-states on CPU with errors in powersaving routines. It appears in xen log:
(XEN) Disabling C-states C3 and C6 on Nehalem Processors due to errata

VHD/VDI

  • snapwatchd daemon removed. But VHD tree is coalescing (still not known all cases - at least at the xe sr-scan).
  • 'Base copy' now named after the original disk name. No longer 'base copy' in name-label
  • vhd coalescing now never remove last base copy (even there is no snapshots or any other reason for splitting). To reduce space usage create new snapshot, delete it, exec xe sr-scan and wait until 'old' and 'new' base copy (disk content prior snapshot creation) will be merged. Disk still gonna continue to grow (base copy is always read only).

Release Notes Errata

Icon Info.png If you find any errors in the XCP 1.6 release notes, please let us know here and we will fix these in the static html pages.