This fixes NPE caused due to merge conflict fix from the forward merge
commit 562a7db8df and fixes travis test
regression:
=== TestName: test_01_reset_vm_on_reboot | Status : SUCCESS ===
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
* Enable unmanaging guest VMs
* Minor fixes
* Fix stop usage event only if VM is not stopped when unmanaging
* Rename unmanaged VMs manager
* Generate netofferingremove usage event if VM is not stopped
* Generate usage event VM snapshot primary off when unmanaging
This PR fixes an issue where an instance fails to deploy due to a null pointer when using an L2 Guest Network with DefaultL2NetworkOfferingConfigDrive on Xenserver. It also fixes migrating an instance to another host.
This has been tested by:
- Creating an L2 Guest network, using DefaultL2NetworkOfferingConfigDrive as the network offering.
- Deploying an instance using the L2 Guest network created.
- Migrating the instance away from the host and back
Repro Steps:
1. Create a VM on host1
2. Make host1 capacity full by deploying multiple VMs
3. Try Dynamic scaling on VM on host1
4. NPE occurs when MS tries to find host to migrate the VM and then scale.
Root cause: VM profile is not initiated properly with serviceoffering before planning for deployment
Solution: Iniate VM profile with serviceoffering and also make sure custom compute parameters are handled
Currently CloudStack is using logging frameworks as log4j and Java util logging, logging wrappers as slf4j and Apache common logging.
Here changes are to made it uniform, using only log4j framework.
Removed Java util logging, slf4j and Apache common logging.
Root cause:
Even though dynamic scaling job is handled in vmworkjob queue which ensures serilizing multiple jobs but the database updating and generating usage events are out of the job queue.
Solution:
Moved all updations into the job queue
Firstly I have tested all the scenarios to check if nothing is broken:
Scaling on a running VM with normal compute offering
Scaling on a stopped VM with normal compute offering
Scaling on a running VM with custom compute offering
Scaling on stopped VM with custom compute offering
Scaling on stopped/running VM between custom compute offering and normal compute offering and combinations among these. Checked if the custom parameters have been populated or deleted accordingly based on the offering to which the VM is scaled
Since this is a corner scenario I could not test the exact point where two usage events are recorded at the same time for two different API calls on same VM.
When expunge a Running vm, vm will be stopped with forcestop=false which does not make sense. we should honor vm.destroy.forcestop in global setting, or always set forcestop=true.
This adds support for JDK11 in CloudStack 4.14+:
- Fixes code to build against JDK11
- Bump to Debian 9 systemvmtemplate with openjdk-11
- Fix Travis to run smoketests against openjdk-11
- Use maven provided jdk11 compatible mysql-connector-java
- Remove old agent init.d scripts
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
* Enable PVLAN support on L2 networks
* Fix prevent null pointer on details
* Add marvin tests
* Fixes from comments
* Fix: missing pvlan type on plugniccommand
* Fix checks on network creation for vlans overlap
* Fix remove prefix from secondary vlan id
* Improve checks on physical network for pvlans
* Fix compatibility with previous pvlan creation
* Fix shared networks backwards pvlan compatibility
* Add ui fix for pvlan type not passed to api
* Add check for isolated vlan id overlap
* Include check for dynamic vlan reserved for secondary vlan
* Fix marvin tests errors
* Fix redundant imports
* Skip marvin test for pvlan if dvswitch is not present
* spelling
Co-authored-by: Andrija Panic <45762285+andrijapanicsb@users.noreply.github.com>
* server: fix resource count of primary storage if some volumes are Expunged but not removed
Steps to reproduce the issue
(1) create a vm and stop it. check resource count of primary storage
(2) download volume. resource count of primary storage is not changed.
(3) expunge the vm, the volume will be Expunged state as there is a volume snapshot on secondary storage. The resource count of primary storage decreased.
(4) update resource count of the account (or domain), the resource count of primary storage is reset to the value in step (2).
* New feature: Add support to destroy/recover volumes
* Add integration test for volume destroy/recover
* marvin: check resource count of more types
* messages translate to JP
* Update messages for CN
* translate message for NL
* fix two issues per Daan's comments
Co-authored-by: Andrija Panic <45762285+andrijapanicsb@users.noreply.github.com>
The VM ingestion feature allows CloudStack to discover, on-board, import existing VMs in an infra. The feature currently works only for VMware, with a hypervisor agnostic framework which may be extended for KVM and XenServer in future.
If the disk size of the vm to be created is greater
than the volume size, then the exception message should
display the numeric value instead of variable name
* marvin: check resource count of more types
* New feature: add flag resource.count.running.vms.only to count resource consumption of only running vms
Stopped VMs do not use CPU/RAM actually.
A new global configuration resource.count.running.vms.only is added to determine whether resource (cpu/memory) of only running vms (including Starting/Stopping) will be taken into calculation of resource consumption.
* Add integration test for resource count of only running vms
When start a vm or migrate a vm (away from a host in host maintenance), cloudstack will check capacity of all hosts and choose one. If there are hundreds of hosts on the platform, it will take some seconds. When cloudstack choose a host and start/migrate vm to it, the resource consumption of the host might have been changed. This normally happens when we start/migrate multiple vms.
It would be better to double check the host capacity when start vm on a host.
This PR includes the fix for cpucore capacity when start/migrate a vm.
* Suqash commits to a single commit and rebase against master
Update marvin tests to use white list
* * Fix marvin test failure
* Add new marvin negative tests cases
* Remove hard-coded hypervisor types in marvin tests
* Fix build error after rebase and add hugepagesless
* Fix readability of python code
* Fix failing test
* Adding cleanup of vms for negative tests
* Bug fixes - change config checks properly and block extraconfig in details
* Trim to compare the keys
* CR comments
* Don't skip extraconfig without exception
Co-authored-by: Boris Stoyanov - a.k.a Bobby <bss.stoyanov@gmail.com>
* 4.13:
only update powerstate if sure it is the latest (#3743)
ui: fix migrate host form no host popup (#3682)
client: jetty session timeout set after server is started (#3658)
Increase DHCP lease time to infinite (#3662)
* Service layer changes for new way of tracking maintanence progress
* Fixes after offline code review
* Fix marvin tests
* Change state name and add documentation
* Fix test
* Fix and add more unit tests for different caseS
* Fix and enhance Marvin Tests
* Fixes for corner cases
* More fixes and logging
* UI fixes
* Some minor changes and reducing VMs on host for more contained tests
* Fixed ssh client auth problem causing test failure
* Code review changes + fixes + some more logging
* Fix flaky tests by adding delays between host states
* Added fetching only enabled hosts for tests
* Make port blocking KVM specific and refactor to handle failure
* Make failing migrations due to tagged host instead of port blocking
* Added additional check for migrating VMs
* Refactor to use single place for methods checking maintenance states
* server: Do NOT cleanup dhcp and dns when stop a vm
According comment in PR #3608, dhcp and dns entries are cleaned up only when a VM is expunged.
Revert part of commit 8fb388e931.
* server: cleanup dns/dhcp entries in removeNic instead of finalizeExpunge
Problem: Users don't know what keys/values to enter for template and VM details.
Root Cause: The feature does not exist that can list possible details and options.
Solution: Based on the possible VM and template details handled by the
codebase, those details were refactored and a list API is introduced
that can return users those details along with possible values. When
users add details now, they will be presented with a list of key details
and their possible options if any.
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
Problem: Not able to configure a sort order for the zones that are listed in various views in the UI.
Root Cause: There is no mechanism to accept sort key for existing zones or UI widget, that would allow to listing zones in the UI in a certain order.
Solution: The order of zones in listed in various views in the UI can now be configured through the newly added “sort_key” field added for the zone. It can be set using updateZone API by providing “sort_key” parameter for a zone, or by reordering the items in the zones list in the UI. UI has been updated to show ordering controls in zones list view. Database changes include updating table “data_center” by adding “sort_key” column (containing integer values and defaults to zero).
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
Feature Specification: https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=95653548
Live storage migration on KVM under these conditions:
From source and destination hosts within the same cluster
From NFS primary storage to NFS cluster-wide primary storage
Source NFS and destination NFS storage mounted on hosts
In order to enable this functionality, database should be updated in order to enable live storage capacibilty for KVM, if previous conditions are met. This is due to existing conflicts between qemu and libvirt versions. This has been tested on CentOS 6 hosts.
Additional notes:
To use this feature set the storage_motion_supported=1 in the hypervisor_capability table for KVM. This is done by default as the feature may not work in some environments, read below.
This feature of online storage+VM migration for KVM will only work with CentOS6 and possible Ubuntu as KVM hosts but not with CentOS7 due to:
https://bugs.centos.org/view.php?id=14026https://bugzilla.redhat.com/show_bug.cgi?id=1219541
On CentOS7 the error we see is: " error: unable to execute QEMU command 'migrate': this feature or command is not currently supported" (reference https://ask.openstack.org/en/question/94186/live-migration-unable-to-execute-qemu-command-migrate/). Reading through various lists looks like the migrate feature with qemu may be available with paid versions of RHEL-EV but not centos7 however this works with CentOS6.
Fix for CentOS 7:
Create repo file on /etc/yum.repos.d/:
[qemu-kvm-rhev]
name=oVirt rebuilds of qemu-kvm-rhev
baseurl=http://resources.ovirt.org/pub/ovirt-3.5/rpm/el7Server/
mirrorlist=http://resources.ovirt.org/pub/yum-repo/mirrorlist-ovirt-3.5-el7Server
enabled=1
skip_if_unavailable=1
gpgcheck=0
yum install qemu-kvm-common-ev-2.3.0-29.1.el7.x86_64 qemu-kvm-ev-2.3.0-29.1.el7.x86_64 qemu-img-ev-2.3.0-29.1.el7.x86_64
Reboot host
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
Since the CloudStack virtual router was redesigned on version 4.6 it has been observed that the DHCP leases file is not persistent across network operations. This causes conflicts on guest VMs static IPs, causing these static IPs to not be renewed by the DHCP server running on isolated and VPC networks' virtual routers (dnsmasq). On stopping or destroying a VM, its dhcp/dns records are not removed from the virtual router causing ghost effects.
Fixes#3272Fixes#3354
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
* Keep connection alive when on maintenance
* Refactor cancel maintenance and unit tests
* Add marvin tests
* Refactor
* Changing the way we get ssh credentials
* Add check on SSH restart and improve marvin tests
* api: add command to list management servers
* api: add number of mangement servers in listInfrastructure command
* ui: add block for mangement servers on infra page
* api name resolution method cleanup
Offerings can co-exist where on does provide Security Grouping in the
network, but other guest Networks have no Security Grouping.
In V(X)LAN isolation environments the L2 separation is handled by V(X)LAN
and protection between Instances is handled by Security Grouping.
There are multiple scenarios possible where one network has Security Grouping
enabled because that is required in that network.
In the other network, but in the same zone it could be a choice to have
Security Grouping disabled and allow all traffic to flow.
Signed-off-by: Wido den Hollander <wido@widodh.nl>
* - Offline VM and Volume migration on Vmware hypervisor hosts
- Also add VM disk consolidation call on successful VM migrations
* Fix indentation of marvin test file and reformat against PEP8
* * Fix few comment typos
* Refactor debug messages to use String.format() when debug log level is enabled.
* Send list of commands returned by hypervisor Guru instead of explicitly selecting the first one
* Fix unhandled NPE during VM migration
* Revert back to distinct event descriptions for VM to host or storage pool migration
* Reformat test_primary_storage file against PEP-8 and Remove unused imports
* Revert back the deprecation messages in the custom StringUtils class to favour the use of the ApacheUtils
With IPv6 we are not using DHCP to allocate addresses, but using
StateLess Address Auto Configuration (SLAAC) a Instance will calculate
it's own address based on the Router Advertisements send out by the
routers in the network.
This Advertisement contains the IPv6 Subnet in use in that subnet and
allows to calculate the stable Address the Instance will obtain based
on it's MAC Address.
The existing code is 'dead code' as it has been written, but was never
used by any production code.
SLAAC only works properly with subnets of exactly 64-bits large.
Signed-off-by: Wido den Hollander <wido@widodh.nl>
This PR adds the possibility to select a checkbox for the parameter bypassvlanoverlapcheck to the ajax request createNetwork. The checkbox was added for Guest Network as well as for the L2 Guest Network. For L2 Guest Network a backend check for the existence of the flag bypassvlanoverlapcheck was added.
This commit allows deploying VMs with a specific IPv4 address.
DirectPodBasedNetworkGuru does not support requesting a custom
IP-Address while creating a new NIC/Instance, throwing the following
error:
Error 530: Does not support custom ip allocation at this time:
NicProfile[0-0-null-null-null
Unknown macro: { "cserrorcode"}
Some use-cases prefer the ability to request the IPv4 address which the
Instance will get.
This implementation adds unit test cases to cover and it was manually
tested in Basic Networking. I can perform more tests if requested.
With earlier work in Basic Networking and the security group provider IPv6 is
supported and we can allow IPv6 to be supplied in networks with SG enabled.
Signed-off-by: Wido den Hollander <wido@widodh.nl>
This force stops old VRs when performing rolling restart with
cleanup=true. This will ensure that VRs are powered off quickly than
wait longer for the normal ACPI shutdown. During testing, it was found
on VMware where VM stops are slow compared to XenServer and KVM.
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
This adds a global setting for admins who may not want the rolling
restart of routers or are seeing any issues around it. In future, this
setting may be removed.
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
These boolean-return methods are named as "getXXX".
Other boolean-return methods are named as "isXXX".
Considering there methods will return boolean values, it should be more clear and consistent to rename them as "isXXX".
(rebase #2602 and #2816)
* Add managed storage pool constraints to MigrateWithVolume API method
* Apply mike's suggestions
* Apply Mike's suggestion in a second review
* Mike's suggestions
* Confused bit
* just executeManagedStorageChecks
* Created methods `executeManagedStorageChecksWhenTargetStoragePoolNotProvided` and `executeManagedStorageChecksWhenTargetStoragePoolProvided`
* improve "executeManagedStorageChecksWhenTargetStoragePoolNotProvided"
* Fix "findVolumesThatWereNotMappedByTheUser" method
* Applu Mike's suggestion to improve "createMappingVolumeAndStoragePool" method
* Unit tests to cover modified code
In integration work for CCS I found that the service call UserVmService.destroyVm(long uuid, boolean expunge) does not honour the expunge flag. I traced it down to the implementation VirtualMachineManagerImpl.destroy(String vmUuid, boolean expunge).
Testing: manual testing so far, testing will pose some crosscutting challanges as the behaviour and implementation are seperated by about five layers of abstraction.
This PR fixes NPE with the provisionCertificateCmd when reconnect is set to True.
Also fixes the following Marvin test failures:
- test_certauthority_root.py
Example: A VM that uses managed storage is stopped. The VM is then started on a different host in the same cluster. The Start operation fails.
To get around this issue, you must either start the VM up on the same host or on a host in a different cluster.
The reason is due to a slightly erroneous check in VolumeOrchestrator.prepare.
To solve this issue, we should be checking if the cluster ID changes, not if the host ID changes.
This introduces a new global setting `vm.configdrive.primarypool.enabled` to toggle creation/hosting of config drive iso files on primary storage, the default will be false causing them to be hosted on secondary storage. The current support is limited from hypervisor resource side and in current implementation limited to `KVM` only. The next big change is that config drive is created at a temporary location by management server and shipped to either KVM or SSVM agent via cmd-answer pattern, the data of which is not logged in logs. This saves us from adding genisoimage dependency on cloudstack-agent pkg.
The APIs to reset ssh public key, password and user-data (via update VM API) requires that VM should be shutdown. Therefore, in the refactoring I removed the case of updation of existing ISO. If there are objections I'll re-put the strategy to detach+attach new config iso as a way of updation. In the refactored implementation, the folder name is changed to lower-cased configdrive. And during VM start, migration or shutdown/removal if primary storage is enable for use, the KVM agent will handle cleanup tasks otherwise SSVM agent will handle them.
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
When a user shuts down their VM from the guest OS (and VM HA is enabled), the VM just powers itself back on. Our environment is on KVM hosts.
CloudStack does not know the difference between a VM failing or being shutdown from within the guest OS.
This is a major pain point for all our users - especially since they don't pay for VMs when they are shutoff. It is not intuitive for end-users to understand why they can't shutdown VMs from within the guest OS. Especially when they all come from (non-cloudstack) VMware and Hyper-V environments where this is not an issue.
However, if a host fails, we need VM HA to still work.
This PR that creates a configuration option "ha.vm.restart.hostup". With this option set to false, if CloudStack sees a VM shutdown out-of-band, but the host it was on is still online, then it won't power the VM back on. The logic is that since the host is online, it was most likely shutdown from the guest OS.
For when a host actually fails, standard VM HA logic takes over and powers on VMs (if they have VM HA enabled) if the host they were on fails.
If that "ha.vm.restart.hostup" option is true (the default to match current functionality), it works like always, and even in-guest shutdowns of VMs causes CloudStack to power back on the VM.
Rohit Yadav
Wei Zhou
Pearl Dsilva
Nicolas Vazquez
Spaceman1984
harikrishna-patnala
dahn
pavanaravapalli
Rakesh
Abhishek Kumar
Anurag Awasthi
Gabriel Beims Bräscher
Anurag Awasthi
Rohit Yadav
Abhishek Kumar
dahn
Wido den Hollander
Malcolm Taylor
Gerd Müller
Gabriel Beims Bräscher
Paul Angus
Kui LIU
René Moser
Frank Maximus
Rafael Weingärtner
lzh3636
Dingane Hlaluku
Mike Tutkowski
Slair1