GuestOS mappings are retrieved from the parent hypervisor version when a minor, patch hypervisor version doesn't exist.
Fixes#8412
Signed-off-by: Abhishek Kumar <abhishek.mrt22@gmail.com>
1. Problem description
In Apache CloudStack (ACS), when a VM is deployed in a host with the KVM hypervisor, an XML file is created in the assigned host, which has a property shares that defines the weight of the VM to access the host CPU. The value of this property has no unit, and it is a relative measure to calculate how much CPU a given VM will have in the host. However, this value has a limit, which depends on the version of cgroup utilized by the host's kernel. The problem lies at the range value of shares that varies between both versions: [2, 264144] for cgroups version 1; and [1, 10000] for cgroups version 2. Currently, ACS calculates the value of shares using Equation 1, presented below, where CPU is the number of cores and speed is the CPU frequency; both specified in the VM's compute offering. Therefore, if a compute offering has, for example, 6 cores at 2 GHz, the shares value will be 12000 and an exception will be thrown by libvirt if the host utilizes cgroup v2. The second version is becoming the default one in current Linux distributions; thus, it is necessary to address this limitation.
Equation 1
shares = CPU * speed
Fixes: #6744
2. Proposed changes
To address the problem described, we propose to apply a scale conversion considering the max shares of the host. Using the same formula currently utilized by ACS, it is possible to calculate the maximum shares of a VM for a given host. In other words, using the number of cores and the nominal speed of the host's CPU as the upper limit of shares allowed to a VM. Then, this value will be scaled to the allowed interval of [1, 10000] of cgroup v2 by using a linear scale conversion.
The VM shares would be calculated as Equation 2, presented below, where VM requested shares is the requested shares value calculated using Equation 1, cgroup upper limit is fixed with a value of 10000 (cgroups v2 upper limit), and host max shares is the maximum shares value of the host, calculated using Equation 1. Using Equation 2, the only case where a VM passes the cgroup v2 limit is when the user requests more resources than the host has, which is not possible with the current implementation of ACS.
Equation 2
shares = (VM requested shares * cgroup upper limit)/host max shares
To implement the proposal, the following APIs will be updated: deployVirtualMachine, migrateVirtualMachine and scaleVirtualMachine. When a VM is being deployed, a new verification will be added to find a suitable host. The max shares of each host will be calculated, and the VM calculated shares will be verified if it does not surpass the host's value. Likewise, the migration of VMs will have a similar new verification. Lastly, the scale of VMs will also have the same verification for the VM's host.
To determine the max shares of a given host, we will use the same equation currently used in ACS for calculating the shares of VMs, presented in Section 1. When Equation 1 is used to determine the maximum shares of a host, CPU is the number of cores of the host, and speed is the nominal CPU speed, i.e., considering the CPU's base frequency.
It is important to note that these changes are only for hosts with the KVM hypervisor using cgroup v2 for now.
Observed a failure to start new virtual machine with PowerFlex storage. Traced it to concurrent VM starts using the same template and the same host to copy. Second mapping attempt failed.
While creating the volume clone from the seeded template in primary storage, adding a lock with the string containing IDs of template, storage pool and destination host avoids the situation of concurrent mapping attempts with the same host.
This removes the conditional logic where comment notest to remove it
after PR #5297 is merged that is applicable for ACS 4.18+. Only when the
global setting is enabled and memory isn't selected, VM snapshot could
be allowed for VMs on KVM that have qemu-guest-agent running.
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
The uprgade from 4.18.1.0 to 4.18.2.0-SNAPSHOT failed with error
```
2023-09-12 16:12:19,003 INFO [c.c.u.DatabaseUpgradeChecker] (main:null) (logid:) DB version = 4.18.1.0 Code Version = 4.18.2.0
2023-09-12 16:12:19,004 INFO [c.c.u.DatabaseUpgradeChecker] (main:null) (logid:) Database upgrade must be performed from 4.18.1.0 to 4.18.2.0
2023-09-12 16:12:19,036 DEBUG [c.c.u.DatabaseUpgradeChecker] (main:null) (logid:) Running upgrade Upgrade41800to41810 to upgrade from 4.18.0.0-4.18.1.0 to 4.18.1.0
...
2023-09-12 16:12:19,041 DEBUG [c.c.u.d.ScriptRunner] (main:null) (logid:) -- Schema upgrade from 4.18.0.0 to 4.18.1.0
...
2023-09-12 16:12:21,602 DEBUG [c.c.u.d.DatabaseAccessObject] (main:null) (logid:) Statement: CREATE INDEX i_cluster_details__name on cluster_details (name)
2023-09-12 16:12:21,663 DEBUG [c.c.u.d.DatabaseAccessObject] (main:null) (logid:) Created index i_cluster_details__name
2023-09-12 16:12:21,673 DEBUG [c.c.u.d.T.Transaction] (main:null) (logid:) Rolling back the transaction: Time = 2632 Name = Upgrade; called by -TransactionLegacy.rollback:888-TransactionLegacy.removeUpTo:831-TransactionLegacy.close:655-TransactionContextInterceptor.invoke:36-ReflectiveMethodInvocation.proceed:175-ExposeInvocationInterceptor.invoke:97-ReflectiveMethodInvocation.proceed:186-JdkDynamicAopProxy.invoke:215-$Proxy30.persist:-1-DatabaseUpgradeChecker.upgrade:319-DatabaseUpgradeChecker.check:403-CloudStackExtendedLifeCycle.checkIntegrity:64
```
It succeeded with this change.
This fixes the following cases in which Solidfire storage integration
caused issues when using Solidfire datadisks with VMware:
1. Take Volume Snapshot of Solidfire data disk
2. Delete an active Instance with Solidfire data disk attached
3. Attach used existing Solidfire data disk to a running/stopped VM
4. Stop and Start an instance with Solidfire data disks attached
5. Expand disk by resizing Solidfire data disk by providing size
6. Expand disk by changing disk offering for the Solidfire data disk
Additional changes:
- Use VMFS6 as managed datastore type if the host supports
- Refactor detection and splitting of managed storage ds name in storage
processor
- Restrict storage rescanning for managed datastore when resizing
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
* VMware: add support for 8.0b (8.0.0.2)
* VMware 8: add new guest os mappings in VirtualMachineGuestOsIdentifier
The full list can be found at https://developer.vmware.com/apis/1355/vsphere
* VMware: get guest os mappings of parent version
* VMware8: remove guest os mappings for 8.0.0.2
* VMware8: fix code smells
* vmware: remove annotations in VmwareVmImplementerTest which caused 0.0% code coverage
* VMware8: add a unit test case
* VMware: add support for 8.0c (8.0.0.3)
* VMware8: move to CloudStackVersion.getVMwareParentVersion
* VMware: add support for 8.0u1 (8.0.1.0)
* Copy engine/schema/src/main/java/com/cloud/upgrade/GuestOsMapper.java from PR 6979
* Copy engine/schema/src/main/java/com/cloud/storage/dao/GuestOSHypervisorDao.java from PR 6979
* VMware: ignore the last number in VMware versions
* VMware: copy guest os mapping from 8.0 to 8.0.1
* VMware: add unit tests in VmwareVmImplementerTest.java
* Copy engine/schema/src/test/java/com/cloud/upgrade/GuestOsMapperTest.java from PR 6979
* VMware8: retry vm poweron if fails due to exception "File system specific implementation of Ioctl[file] failed"
This fixes a weird issue on vmware8. When power on a vm, sometimes it fails due to error
2023-04-27 07:04:43,207 ERROR [c.c.h.v.r.VmwareResource] (DirectAgent-442:ctx-cdd42b03 10.0.32.133, job-105/job-106, cmd: StartCommand) (logid:8a24a607) StartCommand failed due to [Exception: java.lang.RuntimeException
Message: File system specific implementation of Ioctl[file] failed
].
java.lang.RuntimeException: File system specific implementation of Ioctl[file] failed
at com.cloud.hypervisor.vmware.util.VmwareClient.waitForTask(VmwareClient.java:426)
at com.cloud.hypervisor.vmware.mo.VirtualMachineMO.powerOn(VirtualMachineMO.java:288)
in vmware.log on ESXi host, it shows
2023-04-27T09:20:41.713Z In(05)+ vmx - Power on failure messages: File system specific implementation of Ioctl[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of Ioctl[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of LookupAndOpen[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of Ioctl[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - File system specific implementation of Ioctl[file] failed
2023-04-27T09:20:41.713Z In(05)+ vmx - Failed to lock the file
2023-04-27T09:20:41.713Z In(05)+ vmx - Cannot open the disk '/vmfs/volumes/7b29c876-ac102328/i-2-167-VM/ROOT-167.vmdk' or one of the snapshot disks it depends on.
2023-04-27T09:20:41.713Z In(05)+ vmx - Module 'Disk' power on failed.
2023-04-27T09:20:41.713Z In(05)+ vmx - Failed to start the virtual machine.
There is a KB article for it, but I still do not know why and how to fix it.
https://kb.vmware.com/s/article/1004232
* VMware: extract to method powerOnVM
* vmware: fix mistake in logs
* vmware8: use curl instead of wget to fix test failures
Traceback (most recent call last):
File "/root/test_internal_lb.py", line 555, in test_01_internallb_roundrobin_1VPC_3VM_HTTP_port80
self.execute_internallb_roundrobin_tests(vpc_offering)
File "/root/test_internal_lb.py", line 641, in execute_internallb_roundrobin_tests
client_vm, applb.sourceipaddress, max_http_requests)
File "/root/test_internal_lb.py", line 497, in run_ssh_test_accross_hosts
(e, clienthost.public_ip))
AssertionError: list index out of range: SSH failed for VM with IP Address: 10.0.52.187
and
sshClient: DEBUG: {Cmd: /usr/bin/wget -T3 -qO- --user=admin --password=password http://10.1.2.253:8081/admin?stats via Host: 10.0.52.188} {returns: ["/usr/bin/wget: '/usr/lib/libpcre.so.1' is not an ELF file", "/usr/bin/wget: can't load library 'libpcre.so.1'"]}
* VMware: correct guest OS names in hypervisor mappings for VMware 8.0
el9 and variants were introduced by https://github.com/apache/cloudstack/pull/7059
they are supported with guest os identifiers since VMware 8.0
see https://vdc-repo.vmware.com/vmwb-repository/dcr-public/c476b64b-c93c-4b21-9d76-be14da0148f9/04ca12ad-59b9-4e1c-8232-fd3d4276e52c/SDK/vsphere-ws/docs/ReferenceGuide/vim.vm.GuestOsDescriptor.GuestOsIdentifier.html
* VMware: add Ubuntu 20.04 and 22.04 support for vmware 7.0+
* PR7380: only add guest os mappings for Ubuntu 20.04
* PR7380: Correct RHEL9 guest os names and others for VMware 8.0
* PR7380: correct guest os names on 8.0.0.1 as well
* PR7380: remove Windows 12 and Windows Server 2025 which are not released yet
CPU cap limitation was enabled as part of
https://github.com/apache/cloudstack/pull/6420 that changes behaviour
for existing environments. The CPU cap limitation on KVM causes
systemvms to not start or be really slow in nested and virtualised
environments.
Signed-off-by: Rohit Yadav <rohit.yadav@shapeblue.com>
When deploying a VM is failed during the allocation process it may leave the resources that have been already allocated before the failure. They will get removed from the database as the whole code block is wrapped inside a transaction twice but the server would not inform the network or storage plugins to clean up the allocated resources.
This PR removes Transactions during VM allocation which results in the allocated VM and its resource records being persisted in DB even during failures. When failure is encountered VM is moved to Error state. This helps VM and its resources to be properly deallocated when it is expunged either by a server task such as ExpungeTask or during manual expunge.
Signed-off-by: Abhishek Kumar <abhishek.mrt22@gmail.com>
Fixes case of account/domain having negative storage count when there is a volume size difference while deploying volumes on certain stroages. In case of Powerflex volume size results in a multiple of 8. If user deploys a volume of 12GB it will result in 16GB in size. But currently CloudStack will not update resource count after deployment.
Signed-off-by: Abhishek Kumar <abhishek.mrt22@gmail.com>
Supported Virtual machine operations:
- live migration of VM to another host
- virtual machine snapshots (group snapshot without memory)
- revert VM snapshot
- delete VM snapshot
Supported Volume operations:
- attach/detach volume
- live migrate volume between two StorPool primary storages
- volume snapshot
- delete snapshot
- revert snapshot
* Live storage migration of volume in scaleIO within same storage scaleio cluster
* Added migrate command
* Recent changes of migration across clusters
* Fixed uuid
* recent changes
* Pivot changes
* working blockcopy api in libvirt
* Checking block copy status
* Formatting code
* Fixed failures
* code refactoring and some changes
* Removed unused methods
* removed unused imports
* Unit tests to check if volume belongs to same or different storage scaleio cluster
* Unit tests for volume livemigration in ScaleIOPrimaryDataStoreDriver
* Fixed offline volume migration case and allowed encrypted volume migration
* Added more integration tests
* Support for migration of encrypted volumes across different scaleio clusters
* Fix UI notifications for migrate volume
* Data volume offline migration: save encryption details to destination volume entry
* Offline storage migration for scaleio encrypted volumes
* Allow multiple Volumes to be migrated with migrateVirtualMachineWithVolume API
* Removed unused unittests
* Removed duplicate keys in migrate volume vue file
* Fix Unit tests
* Add volume secrets if does not exists during volume migrations. secrets are getting cleared on package upgrades.
* Fix secret UUID for encrypted volume migration
* Added a null check for secret before removing
* Added more unit tests
* Fixed passphrase check
* Add image options to the encypted volume conversion
In troubleshooting ops issues we see logs like:
Maximum domain resource limits of Type 'user_vm' for Domain Id = 763 is exceeded: Domain Resource Limit = (1 bytes) 1, Current Domain Resource Amount = (0 bytes) 0, Requested Resource Amount = (1 bytes) 1."
However there is one missing value (currentResourceReservation) that is used in the calculation of limit check but it is not logged, which leads to confusion. Above we see we are using “0” and requested 1, with our limit being 1, but was rejected. Without logging all the values used in the calculation we don’t understand why it failed.
Additionally, if we had this log above it would be clearer that a second bug is occurring. When we query for domain level resource reservations in “getDomainReservation” the actual SearchBuilder is the listAccountAndTypeSearch, not the listDomainAndTypeSearch. As a result, when we call getDomainReservation the query returns any outstanding domain reservation for any account, as domain ID is not a valid filter for the account search.
This PR:
Increases detailed information in log for checking resource limit to include reservations information for functions: checkDomainResourceLimit() and checkAccountResourceLimit
Fixes getDomainReservation() to use listDomainAndTypeSearch instead of listAccountAndTypeSearch
Co-authored-by: Oscar Sandoval <osandovalocana@apple.com>
