Some issues with managed/unmanaged VMs on KVM

[slavkap]

1. There is an issue while importing a VM with a Basic zone network:

• if you chose to set a random IP address:

2024-01-03 16:51:26,068 ERROR [o.a.c.v.UnmanagedVMsManagerImpl] (API-Job-Executor-10:ctx-bcae51d0 job-864 ctx-e872b0fb) (logid:016f507c) Failed to import NICs while importing vm: i-2-58-VM
com.cloud.exception.InsufficientVirtualNetworkCapacityException: Unable to acquire Guest IP  address for network Network {"id": 204, "name": "defaultGuestNetwork", "uuid": "2c6b7a88-939a-42b1-8a5e-d9641071bff8",
 "networkofferingid": 7}Scope=interface com.cloud.dc.DataCenter; id=1
        at org.apache.cloudstack.engine.orchestration.NetworkOrchestrator.importNic(NetworkOrchestrator.java:4582)

• if you use a predetermined IP address the import fails with

2024-01-03 16:52:49,765 DEBUG [o.a.c.f.j.i.AsyncJobManagerImpl] (API-Job-Executor-11:ctx-a94ebd23 job-865) (logid:0f522606) Complete async job-865, jobStatus: FAILED, resultCode: 530, result: org.apache.cloudsta
ck.api.response.ExceptionResponse/null/{"uuidList":[],"errorcode":"530","errortext":"IP address 10.2.32.49 for NIC(ID: 0) is not available in network(ID: 2c6b7a88-939a-42b1-8a5e-d9641071bff8)"}

2. If the storage pool is of type block for the volumes of an unmanaged VM, CS cannot find the pools. The select DB query tries to list by path and host_address which are taken from the libvirt XML disk definition and are not correct (for example it gets the volume path /dev/storpool-byid/bgu4.b.8wm instead of the actual primary storage path /dev/storpool-byid/. Also StorPool for example doesn't have a host described in the XML definition). Probably this issue will appear with similar storage plugins.

3. Issue with VMs with RBD storage pool. You can unmanage a VM on Ceph storage but the listing of unmanaged VMs fails because libvirt cannot access the RBD volumes.

2024-01-04 11:56:55,383 DEBUG [utils.script.Script] (agentRequest-Handler-2:null) (logid:43ccc964) Executing command [qemu-img info -U cloudstack/d3638e01-955c-4702-8433-b62fe4bc680a ].
2024-01-04 11:56:55,388 WARN  [utils.script.Script] (agentRequest-Handler-2:null) (logid:43ccc964) Execution of process [2868475] for command [qemu-img info -U cloudstack/d3638e01-955c-4702-8433-b62fe4bc680a ] f
ailed.
2024-01-04 11:56:55,388 DEBUG [utils.script.Script] (agentRequest-Handler-2:null) (logid:43ccc964) Exit value of process [2868475] for command [qemu-img info -U cloudstack/d3638e01-955c-4702-8433-b62fe4bc680a ] 
is [1].
2024-01-04 11:56:55,389 WARN  [utils.script.Script] (agentRequest-Handler-2:null) (logid:43ccc964) Process [2868475] for command [qemu-img info -U cloudstack/d3638e01-955c-4702-8433-b62fe4bc680a ] encountered th
e error: [qemu-img: Could not open 'cloudstack/d3638e01-955c-4702-8433-b62fe4bc680a': Could not open 'cloudstack/d3638e01-955c-4702-8433-b62fe4bc680a': No such file or directory].
2024-01-04 11:56:55,389 INFO  [resource.wrapper.LibvirtGetUnmanagedInstancesCommandWrapper] (agentRequest-Handler-2:null) (logid:43ccc964) Unable to retrieve unmanaged instance info. org.apache.cloudstack.utils.
qemu.QemuImgException: qemu-img: Could not open 'cloudstack/d3638e01-955c-4702-8433-b62fe4bc680a': Could not open 'cloudstack/d3638e01-955c-4702-8433-b62fe4bc680a': No such file or directory
java.lang.RuntimeException: org.apache.cloudstack.utils.qemu.QemuImgException: qemu-img: Could not open 'cloudstack/d3638e01-955c-4702-8433-b62fe4bc680a': Could not open 'cloudstack/d3638e01-955c-4702-8433-b62fe
4bc680a': No such file or directory
        at com.cloud.hypervisor.kvm.resource.wrapper.LibvirtGetUnmanagedInstancesCommandWrapper.getUnmanagedInstanceDisks(LibvirtGetUnmanagedInstancesCommandWrapper.java:191)
        at com.cloud.hypervisor.kvm.resource.wrapper.LibvirtGetUnmanagedInstancesCommandWrapper.getUnmanagedInstance(LibvirtGetUnmanagedInstancesCommandWrapper.java:135)

ISSUE TYPE

• Bug Report

CLOUDSTACK VERSION

tested with https://github.com/apache/cloudstack/pull/8433

OS / ENVIRONMENT

Rocky linux 8.5

[DaanHoogland]

@slavkap,
Do you consider any of these issues a blocker?
Does it make sense to split the issue in three tickets?

[slavkap]

@DaanHoogland, it is not a blocker except if somebody uses the Basic network (the first issue).
2 and 3 probably can be fixed first with some limitation in the code that it works only for NFS and shared storage

[shwstppr]

@DaanHoogland @slavkap I don't think the issues should be considered blockers. KVM ingestion has been added recently and maybe we can document some of the unsupported cases. It would be great if we could have fixes for these though
@slavkap do you plan to fix any? 😄
@nvazquez any suggestions as you've recently worked on it?

[slavkap]

@shwstppr, yes, I will work on the second issue and probably will try to fix the problem with Ceph. For the first one, I need to understand the whole feature and right now I don't have much time :(

[DaanHoogland]

NOTE; that the second item of point 1 is still an issue. I am working on it in #8488 . Not sure if we should keep this issue open for that, but it is not completely solved yet.