CloudStack Q&A
Questions tied to Apache CloudStack behaviour and how CMP uses ACS APIs. Feature guides: CloudStack features, Networks.
Can clients set their own password on instance create or password reset?
No — not through a native CloudStack parameter that CMP can pass.
CloudStack does not support end-user-chosen passwords for:
- VM deploy
- Reset password
Standard ACS workflow:
- CloudStack generates a random password
- Password is injected into the guest (password-enabled template + User Data / VR services)
- Generated password is returned via API and shown in CMP
CMP therefore displays the auto-generated password from CloudStack. There is no supported ACS API field to supply a custom password at deploy or reset time.
Ensure templates are password-enabled and network offerings include User Data — see Virtual Router/VPC — User Data.
Plans are in the admin panel but missing on Create Instance. Why?
Check all of the following:
- Package status — must be Active
- Zone and OS family — Create Instance filters by selected zone and OS; Windows/Linux-restricted packages only appear for the matching family
- Plan categories — if plan categories are enabled, every visible package must have a plan category; packages without one are hidden
- Compute categories — if compute categories are configured, packages must be assigned; unassigned packages stay hidden
We changed the rate card offering (40 GB → 60 GB disk, same price). Do existing VMs upgrade?
No. Updating the rate card package / offering only affects new orders.
Existing VMs keep the offering they were provisioned with. To grow disk on running VMs, use the VM upgrade (or resize) flow per instance so ACS and CMP update the associated offering and disk.
Why is no IP listed under Isolated Network → IP Addresses?
Expected when the VM only has a private IP and CMP is reusing Source NAT.
- The first public IP of an isolated network is the Source NAT IP
- CMP reuses it with port forwarding for the first public-access VM
- That SNAT IP is not listed as a separately acquired IP under IP Addresses in that empty/private-only case
When the customer requests public IP on create:
- If SNAT is free → reuse SNAT + port forwarding
- If SNAT is already tied to a VM → acquire a new public IP
Details: Isolated Network — Source NAT reuse.
What does “No destination found for a deployment” mean?
Example error from CloudStack (also shown in CMP when start/deploy fails):
No destination found for a deployment for VM instance
{"id":871,"instanceName":"i-2-871-QA","state":"Stopped",...}
Meaning: CloudStack’s deployment planner could not find any host that meets all placement rules for that VM. It evaluated candidates in the cluster/zone and rejected every one — so start, deploy, or migrate fails.
This is almost always an infrastructure / placement problem in CloudStack, not a CMP package misconfiguration (unless tags on the service/disk offering do not match any host or storage).
Common causes
| # | Cause | What to check |
|---|---|---|
| 1 | Host capacity exhausted (most common) | CPU / memory / storage allocation on hosts — look for InsufficientServerCapacityException in logs |
| 2 | Primary storage unavailable | Primary storage Up, connected, not in maintenance; pool reachable from candidate hosts |
| 3 | Host or cluster maintenance / down | Host status Up + Enabled — not Maintenance, Disconnected, or Alert |
| 4 | Hypervisor cannot see the disk pool | KVM: virsh pool-list, NFS mounts, local disk space |
| 5 | Affinity / anti-affinity | Affinity groups blocking co-location or separation |
| 6 | Dedicated resources | Dedicated account / domain / cluster / host limits where the VM can run |
| 7 | Host tags | Service offering Host tags must match at least one eligible host (for example gpu=true) |
| 8 | Storage tags | Disk offering Storage tags must match a primary pool (for example ssd vs nvme) |
| 9 | Local storage + original host gone | Root disk on Local storage only exists on one host — if that host is down, no destination exists |
| 10 | Agent disconnected | cloudstack-agent running; host shows Up in CloudStack |
For a Stopped VM that fails to start, start with capacity, host/storage tags, storage accessibility, maintenance, and (if used) local storage on a missing host.
How to debug
1. Confirm host and storage health (UI)
- Infrastructure → Hosts — capacity, status, tags
- Infrastructure → Primary Storage — Up, not in maintenance
- Infrastructure → Dedicated Resources / Compute → Affinity Groups if used
Or via API / CloudMonkey: list hosts, list storagepools.
2. Read why each host was skipped (best step)
On the management server:
grep -i -A50 -B50 "No destination found for a deployment" \
/var/log/cloudstack/management/management-server.log
Useful related searches:
grep -i "Insufficient" /var/log/cloudstack/management/management-server.log
grep -i "DeploymentPlanner" /var/log/cloudstack/management/management-server.log
grep -A20 -B20 "871" /var/log/cloudstack/management/management-server.log # use your VM id
A few lines above No destination found, CloudStack usually logs why hosts were rejected, for example:
- insufficient memory / CPU
- host tag mismatch
- storage inaccessible
- affinity / dedication constraints
That line is the real root cause.
3. Optional DB checks
SELECT id, name, state, host_id, last_host_id
FROM vm_instance
WHERE id = 871;
SELECT id, name, pool_type FROM storage_pool;
If the VM’s root volume is on pool_type = Local and last_host_id is unavailable, placement will fail until that host returns or the disk is on shared storage.
4. KVM host (if applicable)
systemctl status cloudstack-agent
virsh pool-list
mount | grep nfs
df -h
How to resolve
Fix the underlying rejection reason from the logs, then retry start/deploy from CMP or CloudStack:
| If logs show… | Typical fix |
|---|---|
| Insufficient CPU/memory | Free capacity, add hosts, or use a smaller offering |
| Tag mismatch | Align service/disk offering tags with host/storage tags, or clear unnecessary tags |
| Storage down / inaccessible | Repair primary storage connectivity; take pool out of maintenance |
| Host in maintenance | Cancel maintenance or enable another host in the cluster |
| Local disk + dead host | Restore the original host, or restore/migrate data per your DR process (local disks do not float to other hosts) |
| Affinity / dedication | Adjust affinity groups or dedicated resource scope |
When customers see this on Start Instance, treat it as CloudStack capacity/placement. Check ACS first; changing CMP rate cards will not fix a full cluster or a tag mismatch.
How to change the region (zone) name?
Same as other orchestrators in CMP: Settings → Orchestrator → Zones. See All orchestrators — region name.