Skip to content
Commits on Sep 10, 2015
  1. @johnelse

    Merge pull request #2407 from johnelse/typos

    johnelse committed Sep 10, 2015
    Fix some logging typos
  2. @johnelse

    Fix some logging typos

    johnelse committed Sep 10, 2015
    Signed-off-by: John Else <john.else@citrix.com>
  3. @johnelse

    Merge pull request #2405 from johnelse/ca-182869

    johnelse committed Sep 10, 2015
    CA-182869: Revert "CAR-1744: Escape the value strings passed between xe command and CLI server"
  4. @johnelse

    CA-182869: Revert "CAR-1744: Escape the value strings passed between …

    johnelse committed Sep 10, 2015
    …xe command and CLI server"
    
    This reverts commit ed6840d.
    
    This patch was unescaping escaped values on the CLI, causing problems
    with the XenRT RBAC tests.
    
    Signed-off-by: John Else <john.else@citrix.com>
Commits on Sep 9, 2015
  1. @johnelse

    Merge pull request #2403 from johnelse/revert-rrdd

    johnelse committed Sep 9, 2015
    CA-177513: Revert changes to remove master/slave/knowledge from rrdd
  2. @zli

    Merge pull request #2308 from zli/CAR-1744

    zli committed Sep 9, 2015
    CAR-1744: add support for running in-guest user scripts
  3. @zli

    CAR-1744: better logging for VM.call_plugin

    zli committed Jul 2, 2015
    As the args are for the plugin rather than XAPI, it's better to keep a record
    for debugging purpose. This is also a requirement from REQ-103.
    
    Signed-off-by: Zheng Li <dev@zheng.li>
  4. @zli

    CAR-1744: add call_plugin to the vm_operation_table for better error …

    zli committed Jun 22, 2015
    …message
    
    Signed-off-by: Zheng Li <dev@zheng.li>
  5. @zli

    CAR-1744: VM.call_plugin should fail immediately if other operation i…

    zli committed Jun 22, 2015
    …n progress
    
    This is not a normal XAPI operation but an external request. By design, we're
    required to fail it quickly and tell the caller to retry later instead of
    keeping it waiting indefinitely.
    
    Signed-off-by: Zheng Li <dev@zheng.li>
  6. @zli

    CAR-1744: Add a fail_immediately policy for message_forwarding's retr…

    zli committed Jun 22, 2015
    …y logic
    
    In some cases we'd like to fail immediately (and to notifiy the caller/user) if
    other operation is in progress.
    
    Instead of duplicating existing code in higher level functions such as
    with_vm_operation for this special case (no retry), we simply add this special
    policy and expose the policy parameter to the caller.
    
    Note that even though it's called fail_immediately, we deliberately keep a
    minimal interval coverage of 0-3 seconds. Two benefits:
    
     * This could avoid staggering by accident
     * This could help to avoid DDOS
    
    Signed-off-by: Zheng Li <dev@zheng.li>
  7. @zli

    CAR-1744: add throttling mechanism to VM.call_plugin

    zli committed Jun 16, 2015
    Signed-off-by: Zheng Li <dev@zheng.li>
  8. @zli

    CAR-1744: add pool wide flag to disable guest agent run-script by def…

    zli committed Jun 16, 2015
    …ault
    
    Signed-off-by: Zheng Li <dev@zheng.li>
  9. @zli

    CAR-1744: Make sure that call_plugin can be aborted by hard shutdown/…

    zli committed Sep 4, 2015
    …reboot
    
    Signed-off-by: Zheng Li <dev@zheng.li>
  10. @zli

    CAR-1744: run-script plugin implementation

    zli committed Sep 4, 2015
    Signed-off-by: Zheng Li <dev@zheng.li>
  11. @zli

    CAR-1744: function that supports getting xenopsd task result while sy…

    zli committed Sep 4, 2015
    …ncing
    
    The current sync_task implementation always ignores xenopsd task result. Here
    we add a new function which will get the result while syncing is done.
    
    Signed-off-by: Zheng Li <dev@zheng.li>
  12. @zli

    CAR-1744: Escape the value strings passed between xe command and CLI …

    zli committed Jun 8, 2015
    …server
    
    In some cases, we allow to pass file contents as arguments. They might become
    mangled without escaping.
    
    Signed-off-by: Zheng Li <dev@zheng.li>
  13. @johnelse
  14. @johnelse
  15. @johnelse
  16. @johnelse

    Revert "CP-12141: Remove all master/slave knowledge from xcp-rrdd"

    johnelse committed Sep 9, 2015
    This reverts commit a873658.
    
    Conflicts:
    	ocaml/xapi/xapi_vm.ml
  17. @johnelse
  18. @johnelse

    Revert "CA-177513: Push RRDs to slaves correctly"

    johnelse committed Sep 9, 2015
    This reverts commit b7237f6.
  19. @jonludlam

    Merge pull request #2400 from thomassa/shut-shut

    jonludlam committed Sep 9, 2015
    CA-182595 VM.hard_shutdown: use subtask for cancelling
  20. Merge pull request #2402 from djs55/release.1.9.80

    committed Sep 9, 2015
    Prepare to release 1.9.80
  21. Prepare to release 1.9.80

    committed Sep 9, 2015
    Signed-off-by: David Scott <dave@recoil.org>
Commits on Sep 8, 2015
  1. @thomassa

    Make call_slave_with_session simpler and safer

    thomassa committed Sep 7, 2015
    Message_forwarding.call_slave_with_session
    
    Overall the effect is the same, but we save creating and destroying a
    task, the code is easier to understand, and as soon as the new session
    exists, everything else is done with "finally destroy the session".
    
    Signed-off-by: Thomas Sanders <thomas.sanders@citrix.com>
  2. @thomassa

    CA-182595 VM.hard_shutdown: use subtask for cancelling

    thomassa committed Sep 1, 2015
    Also same for VM.hard_reboot.
    
    When Message_forwarding.VM.hard_shutdown cancels other tasks such as
    clean_shutdown, now it uses a fresh subtask for each cancellation,
    instead of doing the cancellation more directly.
    
    This fixes a bug: if a clean_shutdown task was in progress on a VM on
    a slave, then VM.hard_shutdown invoked via "xe vm-shutdown --force"
    would fail after cancelling the clean_shutdown. This was because when
    the call to the slave (to cancel the clean_shutdown) completes, it
    marks the task running it as completed successfully. Since this task
    was the main task for the hard_shutdown, the waiter in cli_operations
    noticed and destroyed the task, logging out the corresponding xapi
    session and telling xe that the command had succeeded. Therefore when
    Message_forwarding.VM.hard_shutdown went on to its next step and
    tried to tell the slave to shut down the VM, it was using an expired
    xapi session so it failed.
    
    Therefore using a new subtask for the cancellation fixes the bug.
    
    The same problem was not observed for "xe vm-reboot --force" because
    in Cli_operations.vm_reboot makes a straightforward synchronous call,
    whereas vm_shutdown makes an async call and passes the task to the
    waiter function.  Nonetheless, the task was still being marked
    prematurely as completed, so this commit makes the same change in
    Message_forwarding.VM.hard_reboot as in the corresponding shutdown.
    
    Signed-off-by: Thomas Sanders <thomas.sanders@citrix.com>
Commits on Sep 7, 2015
  1. @johnelse

    Merge pull request #2399 from johnelse/ca-180871-2

    johnelse committed Sep 7, 2015
    CA-180871: Whitelist the rawhba SR
  2. @johnelse

    CA-180871: Whitelist the rawhba SR

    johnelse committed Sep 3, 2015
    Signed-off-by: John Else <john.else@citrix.com>
Commits on Sep 4, 2015
  1. @jonludlam

    Merge pull request #2398 from djs55/ca-181545

    jonludlam committed Sep 4, 2015
    CA-181545: Query the message switch to find which of the whitelisted …
  2. @johnelse

    CA-181545: Query the message switch to find which of the whitelisted …

    johnelse committed with David Scott Sep 4, 2015
    …plugins are running
    
    If we don't do this, xapi will then query the missing plugin and this
    query will hang forever.
    
    Signed-off-by: John Else <john.else@citrix.com>
    Signed-off-by: David Scott <dave.scott@citrix.com>
  3. @johnelse

    Merge pull request #2397 from johnelse/ca-180871-revert

    johnelse committed Sep 4, 2015
    Revert "CA-180871: Whitelist the rawhba SR"
  4. @johnelse

    Revert "CA-180871: Whitelist the rawhba SR"

    johnelse committed Sep 4, 2015
    If a plugin is whitelisted but doesn't actually exist, xapi assumes it
    is a SMAPIv2 plugin and hangs forever trying to query it. This needs
    fixing before we can whitelist rawhba.
    
    This reverts commit a6c4b58.
    
    Signed-off-by: John Else <john.else@citrix.com>
Commits on Sep 3, 2015
  1. @johnelse

    Merge pull request #2395 from johnelse/ca-180871

    johnelse committed Sep 3, 2015
    CA-180871: Whitelist the rawhba SR
  2. @johnelse

    CA-180871: Whitelist the rawhba SR

    johnelse committed Sep 3, 2015
    Signed-off-by: John Else <john.else@citrix.com>
Something went wrong with that request. Please try again.