Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
Commits on May 13, 2013
  1. Set driver domain when plugging a PIF

    authored
    Signed-off-by: Rob Hoes <rob.hoes@citrix.com>
  2. Set VIF backends properly with regards to network driver domains

    authored
    Signed-off-by: Rob Hoes <rob.hoes@citrix.com>
  3. Pass through PCI devices to network driver domains as required

    authored
    Signed-off-by: Rob Hoes <rob.hoes@citrix.com>
  4. Add utility functions for network driver domains to Network module

    authored
    * Network.driver_domain_devs returns the PCI devices that are "managed"
      by a given VM as configured in other-config:managed-pci-uuids.
    * Network.get_driver_domain returns, for a given PIF, the VM that is
      "managing" it (as above). It furthermore behaves correctly for VLANs
      and other higher-level PIFs: e.g. a VLAN PIF is managed by whatever
      VM manages its underlying physical PIF.
    
    Signed-off-by: Rob Hoes <rob.hoes@citrix.com>
  5. networkd: Obtain MAC addresses from sysfs rather than the ip command

    authored
    Signed-off-by: Rob Hoes <rob.hoes@citrix.com>
  6. networkd: Enable use in a driver domain

    authored
    * Add {Interface,Bridge}.set_driver_domain functions to allow the
      client (xapi) to specify which VM owns the interface or bridge
      (by UUID). Networkd keeps track of this in its persistent state.
    * Introduce a forwarding layer for API calls. A call is executed
      locally or sent off to a networkd instance in a driver domain
      depending on the above mentioned driver-domain state.
    * When an API call is forwarded, the IP address of the driver doamin
      is obtained from xenstore.
    * Functions such as Interface.get_all query all known driver domains
      to obtain the required information across the host.
    * We also expose some further utility functions: Interface.{is_vif_front,
      get_pci_bus_path,rename}.
    
    Signed-off-by: Rob Hoes <rob.hoes@citrix.com>
  7. xenopsd: Set permissive flag in pciback for PCI passthru

    authored
    This is necessary for certain network drivers to function in a driver
    domain. Perhaps we need to make it configurable, as I am not sure whether
    this options is safe to be enabled in all cases.
    
    Signed-off-by: Rob Hoes <rob.hoes@citrix.com>
  8. networkd: setup HIMN interface in driver domain

    authored
    * Rename it to "xenapi".
    * Obtain an IP from DHCP.
    * Listen on the IP.
    * Write the IP to xenstore.
    
    Signed-off-by: Rob Hoes <rob.hoes@citrix.com>
Commits on May 10, 2013
  1. @jonludlam

    Merge pull request #1204 from BobBall/ConsoleFix

    jonludlam authored
    Adding console examples to git repo
  2. @mcclurmc

    Merge pull request #1061 from akshayramani/CA-88840

    mcclurmc authored
    CA-88840:User Friendly error message when cross/intrapool migration fails
  3. @mcclurmc

    Merge pull request #1210 from zli/xenserver/PR-1455

    mcclurmc authored
    xenserver/PR-1455
  4. @jonludlam

    Merge pull request #1224 from jonludlam/CA-105002

    jonludlam authored
    CA-105002: Robustify against domains that are unkillable
  5. @jonludlam

    CA-105002: Robustify against domains that are unkillable

    jonludlam authored
    Signed-off-by: Jon Ludlam <jonathan.ludlam@eu.citrix.com>
Commits on May 9, 2013
  1. @jonludlam

    Merge pull request #1219 from thomassa/CA-105080

    jonludlam authored
    CA-105080: just added some comments
  2. @jonludlam

    Merge pull request #1220 from mcclurmc/ca106034-xenstore-watches-master

    jonludlam authored
    CA-106034: update device_watches when domain appears/disappears
  3. CA-106034: update device_watches when domain appears/disappears

    Mike McClurg authored
    When a domain appears, we need to initialise the entry in device_watches for
    that domid. When a domain disappears, we need to remove the entry in
    device_watches for that domid.
    
    This was overlooked in the refactoring of CA-102259 -- the old semantics of
    add_domU_watches and remove_domU_watches was not preserved as the watches map
    was split into a set of domain watches and a map of device_watches.
    
    Acked-by: Jonathan Davies <Jonathan.Davies@eu.citrix.com>
    Signed-off-by: Mike McClurg <mike.mcclurg@citrix.com>
  4. @thomassa

    CA-105080: just added some comments

    thomassa authored
    Signed-off-by: Thomas Sanders <thomas.sanders@citrix.com>
  5. Merge pull request #1184 from zli/xenserver/CA-102759

    authored
    CA-102759: loosen HA restriction on expired hosts
  6. @jonludlam

    Merge pull request #1216 from jonludlam/CA-95053-v2

    jonludlam authored
    CA-95053: VM.start should check if a host has enough memory
Commits on May 8, 2013
  1. @jonludlam

    CA-95053: VM.start should check if a host has enough memory

    jonludlam authored
    Though it should fall back to just trying anyway in case of
    failure as memory checks are notoriously hard to get right.
    
    This patch is based on a similar one by Jerome Maloberti.
    
    Signed-off-by: Jon Ludlam <jonathan.ludlam@eu.citrix.com>
  2. @jonludlam

    Merge pull request #1215 from jonludlam/CA-105003

    jonludlam authored
    CA-105003: Handle exceptions in get_shadow_allocation.
  3. @jonludlam

    CA-105003: Handle exceptions in get_shadow_allocation.

    jonludlam authored
    This fails when the VM has no VCPUs, which has been observed in
    the wild. Also contains a workaround for CA-104562 by preventing
    bizarrely large values being set.
    
    Signed-off-by: Jon Ludlam <jonathan.ludlam@eu.citrix.com>
  4. @mcclurmc
  5. @mcclurmc

    Merge pull request #1214 from akshayramani/CA-104869

    mcclurmc authored
    CA-104869: Generation-id value incorrect when prepended with leading zer...
  6. @jonludlam

    Merge pull request #1199 from jonludlam/SCTX-1334

    jonludlam authored
    SCTX-1334 - reviewed by @djs55
  7. @jonludlam

    Merge pull request #1212 from jonludlam/CA-87689

    jonludlam authored
    Clean up the snapshot VDI if SXM fails
  8. @akshayramani

    CA-104869: Generation-id value incorrect when prepended with leading …

    akshayramani authored
    …zeros.
    
    Signed-off-by: Akshay Ramani <akshay.ramani@citrix.com>
  9. @jonludlam

    Clean up the snapshot VDI if SXM fails

    jonludlam authored
    Signed-off-by: Jon Ludlam <jonathan.ludlam@eu.citrix.com>
  10. @jonludlam

    Raise the correct error if a xenops task is cancelled

    jonludlam authored
    Signed-off-by: Jon Ludlam <jonathan.ludlam@eu.citrix.com>
Commits on May 7, 2013
  1. @jonludlam

    SCTX-1334: Add exception handlers

    jonludlam authored
    In response to code review with djs55
    
    Signed-off-by: Jon Ludlam <jonathan.ludlam@eu.citrix.com>
  2. @jonludlam

    Merge pull request #1206 from mcclurmc/ca105098-augusta-vs-clearwater

    jonludlam authored
    CA-105098: GenID field was rel_clearwater, not rel_augusta
  3. @jonludlam

    Merge pull request #1209 from vbmithr/master

    jonludlam authored
    CP-4666 improvements to xapi_vm_lifecycle.ml
  4. @jonludlam

    CA-105793: Don't ignore the shutdown_delay parameter.

    jonludlam authored
    Signed-off-by: Jon Ludlam <jonathan.ludlam@eu.citrix.com>
  5. @jonludlam

    SCTX-1334: Ensure that a force shutdown cancels any clean_shutdown.

    jonludlam authored
    Signed-off-by: Jon Ludlam <jonathan.ludlam@eu.citrix.com>
  6. @zli

    PR-1455: Update pool's other-config:mail-min-priority to the new aler…

    zli authored
    …ts grading system
    
    Users were allowed to set the mail-min-priority to arbitray value before, so there isn't a definitive algorithm to map the setting to the new alert grades. This is just one reasonble way making the transformation:
    
    Given previous mail-min-priority = n
    
        n > 10       ->  0
        n = 10       ->  1
        5 < n < 10   ->  2
        n = 5        ->  3
        1 < n < 5    ->  4
        n <= 1       ->  5
        unrecognized ->  remove (reset to default)
    
    Signed-off-by: Zheng Li <zheng.li@eu.citrix.com>
Something went wrong with that request. Please try again.