[
https://issues.apache.org/jira/browse/CLOUDSTACK-6460?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14134040#comment-14134040
]
Simon Weller commented on CLOUDSTACK-6460:
------------------------------------------
darrentang,
Please use the new patch I just uploaded (CLOUDSTACK-6460-darrentang_2.patch).
qemu-img is no longer being used, as the code is now detecting its raw to raw,
so it's using cp instead.
The issue here seems to be that for some reason .raw is being appended onto the
path, and rather than using qemu-img to determine the file type, the code makes
as assumption based on the incorrect file extension that it's a raw file, when
it fact it was a QCOW2 file. This patch should get you going, although really a
hack. The entire code based around this functionality needs to be re-factored,
as it's using lots of (bad) assumptions, and numerous ugly conditional
statements that have been jacked into the code via different developers for new
features.
> Migration of CLVM volumes to another primary storage fail
> ---------------------------------------------------------
>
> Key: CLOUDSTACK-6460
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-6460
> Project: CloudStack
> Issue Type: Bug
> Security Level: Public(Anyone can view this level - this is the
> default.)
> Components: KVM, Volumes
> Affects Versions: 4.2.0, 4.2.1, 4.3.0, 4.4.0
> Environment: KVM clusters with fiber channel SAN storage, CLVM volumes
> Reporter: Salvatore Sciacco
> Attachments: CLOUDSTACK-6460-darrentang.patch,
> CLOUDSTACK-6460-darrentang_2.patch, cloudstack-6460.patch,
> cloudstack-6460_44.patch
>
>
> ACS version: 4.2.1
> Hypervisors: KVM
> Storage pool type: CLVM
> Since we upgraded from 4.1 to 4.2.1 moving volumes to a different primary
> storage pool fail. I've enabled debug on the agents side and I think there is
> a problem with the format type conversion
> Volume on database had format QCOW2
> these are the parameters for the first step (CLVM -> NFS):
> {quote}
> "srcTO":{"org.apache.cloudstack.storage.to.VolumeObjectTO":
> "uuid":"cda46430-52d7-4bf0-b0c2-adfc78dd011c","volumeType":"ROOT","dataStore":{"org.apache.cloudstack.storage.to.PrimaryDataStoreTO":"uuid":"655d6965-b3f3-4118-a970-d50cf6afc365","id":211,"poolType":"CLVM","host":"localhost","path":"/FC10KY1","port":0,"name":"ROOT-4450","size":5368709120,"path":"39a25daf-23a1-4b65-99ac-fb98469ac197","volumeId":5937,"vmName":"i-402-4450-VM","accountId":402,"format":"QCOW2","id":5937,"hypervisorType":"KVM"}
> "destTO":{"org.apache.cloudstack.storage.to.VolumeObjectTO":{"uuid":"cda46430-52d7-4bf0-b0c2-adfc78dd011c","volumeType":"ROOT","dataStore":{"com.cloud.agent.api.to.NfsTO":
>
> "_url":"nfs://192.168.11.6/home/a1iwstack","_role":"Image"},"name":"ROOT-4450","size":5368709120,"path":"volumes/402/5937","volumeId":5937,"vmName":"i-402-4450-VM","accountId":402,"format":"QCOW2","id":5937,"hypervisorType":"KVM"}
> {quote}
> Those commads are translated into the agent:
> {quote}
> DEBUG [utils.script.Script] (agentRequest-Handler-1:null) Executing: qemu-img
> info /dev/FC10KY1/39a25daf-23a1-4b65-99ac-fb98469ac197
> DEBUG [utils.script.Script] (agentRequest-Handler-1:null) Execution is
> successful.
> DEBUG [utils.script.Script] (agentRequest-Handler-1:null) Executing:
> /bin/bash -c cp -f /dev/FC10KY1/39a25daf-23a1-4b65-99ac-fb98469ac197
> /mnt/b8311c72-fe75-3832-98fc-975445028a12/5c713376-c418-478c-8a31-89c4181cb48e.qcow2
>
> {quote}
> With the result that the output file isn't a qcow2 file but a raw partition,
> which in turn make the next step fail.
> (NFS -> CLVM)
> {quote}
> DEBUG [utils.script.Script] (agentRequest-Handler-2:) Executing: qemu-img
> info
> /mnt/b8311c72-fe75-3832-98fc-975445028a12/b9303d8d-cd51-4b6c-a244-43c405df4238.qcow2
>
> DEBUG [utils.script.Script] (agentRequest-Handler-2:) Execution is successful.
> DEBUG [utils.script.Script] (agentRequest-Handler-2:) Executing: qemu-img
> convert -f qcow2 -O raw
> /mnt/b8311c72-fe75-3832-98fc-975445028a12/b9303d8d-cd51-4b6c-a244-43c405df4238.qcow2
> /dev/FCSTORAGE/da162325-467b-4e78-af07-4bad85470d66
> DEBUG [utils.script.Script] (agentRequest-Handler-2:) Exit value is 1
> DEBUG [utils.script.Script] (agentRequest-Handler-2:) qemu-img: Could not
> open
> '/mnt/b8311c72-fe75-3832-98fc-975445028a12/b9303d8d-cd51-4b6c-a244-43c405df4238.qcow2'qemu-img:
> Could not open
> '/mnt/b8311c72-fe75-3832-98fc-975445028a12/b9303d8d-cd51-4b6c-a244-43c405df4238.qcow2'
> ERROR [kvm.storage.LibvirtStorageAdaptor] (agentRequest-Handler-2:) Failed to
> convert
> /mnt/b8311c72-fe75-3832-98fc-975445028a12/b9303d8d-cd51-4b6c-a244-43c405df4238.qcow2
> to /dev/FCSTORAGE/da162325-467b-4e78-af07-4bad85470d66 the error was:
> qemu-img: Could not open
> '/mnt/b8311c72-fe75-3832-98fc-975445028a12/b9303d8d-cd51-4b6c-a244-43c405df4238.qcow2'qemu-img:
> Could not open
> '/mnt/b8311c72-fe75-3832-98fc-975445028a12/b9303d8d-cd51-4b6c-a244-43c405df4238.qcow2'
> {quote}
> If I change on the database the format of the volume to RAW the effect is
> even worse as data is lost in the process!
> These are the parameter for the first step (CLVM => NFS)
> {quote}
> "srcTO":{"org.apache.cloudstack.storage.to.VolumeObjectTO":{"uuid":"cda46430-52d7-4bf0-b0c2-adfc78dd011c","volumeType":"ROOT","dataStore":{"org.apache.cloudstack.storage.to.PrimaryDataStoreTO":{"uuid":"655d6965-b3f3-4118-a970d50cf6afc365","id":211,"poolType":"CLVM","host":"localhost","path":"/FC10KY1","port":0,"name":"ROOT-4450"
> ,"size":5368709120,"path":"39a25daf-23a1-4b65-99ac-fb98469ac197","volumeId":5937,"vmName":"i-4024450VM","accountId":402,"format":"RAW","id":5937,"hypervisorType":"KVM"
> "destTO":{"org.apache.cloudstack.storage.to.VolumeObjectTO":"uuid":"cda46430-52d7-4bf0-b0c2-adfc78dd011c","volumeType":"ROOT","dataStore":{"com.cloud.agent.api.to.NfsTO":
>
> "_url":"nfs://192.168.11.6/home/a1iwstack","_role":"Image"}},"name":"ROOT4450","size":5368709120,"path":"volumes/402/5937","volumeId":5937,"vmName":"i-402-4450-VM","accountId":402,"format":"RAW","id":5937,"hypervisorType":"KVM"}
> {quote}
> this time the output is converted to qcow2!
> {quote}
> DEBUG [utils.script.Script] (agentRequest-Handler-3:null) Executing: qemu-img
> info /dev/FC10KY1/39a25daf-23a1-4b65-99ac-fb98469ac197
> DEBUG [utils.script.Script] (agentRequest-Handler-3:null) Execution is
> successful.
> DEBUG [utils.script.Script] (agentRequest-Handler-3:null) Executing: qemu-img
> convert -f raw -O qcow2 /dev/FC10KY1/39a25daf-23a1-4b65-99ac-fb98469ac197
> /mnt/b8311c72-fe75-3832-98fc-975445028a12/01ab129f-aaf6-4b1a-8e2a-093bee0b811c.raw
>
> {quote}
> and data is lost in the next step (NFS -> CLVM):
> {quote}
> "srcTO":{"org.apache.cloudstack.storage.to.VolumeObjectTO":{"uuid":"cda46430-52d7-4bf0-b0c2-adfc78dd011c","volumeType":"ROOT","dataStore":{"com.cl
> oud.agent.api.to.NfsTO":{"_url":"nfs://192.168.11.6/home/a1iwstack","_role":"Image"}},"name":"ROOT4450","size":5368709120,"path":"volumes/402/5937/01ab129f-aaf6-4b1a-8e2a-093bee0b811c.raw","volumeId":5937,"vmName":"i-402-4450-VM","accountId":402,"format":"RAW","id":5937,"hypervisorType":"KVM"
> "destTO":{"org.apache.cloudstack.storage.to.VolumeObjectTO":{"uuid":"2d684689-efbc-4ce3-95ad-b38e80a1afcf","volumeType":"ROOT","dataStore":{"org.apache.cloudstack.storage.to.PrimaryDataStoreTO":{"uuid":"4d59fde3-4c93-4aa5-8041-fab797b90899","id":214,"poolType":
> "CLVM","host":"localhost","path":"/STOREY1","port":0}},"name":"ROOT-4450","size":5368709120,"path":"39a25daf-23a1-4b65-99ac-fb98469ac197","volumeId":5967,"vmName":"i-402-4450-VM","accountId":402,"format":"RAW","id":5967,"hypervisorType":"KVM"
> DEBUG [utils.script.Script] (agentRequest-Handler-4:null) Executing: qemu-img
> info
> /mnt/b8311c72-fe75-3832-98fc-975445028a12/01ab129f-aaf6-4b1a-8e2a-093bee0b811c.raw
>
> DEBUG [utils.script.Script] (agentRequest-Handler-4:null) Execution is
> successful.
> DEBUG [utils.script.Script] (agentRequest-Handler-4:null) Executing:
> /bin/bash -c cp -f
> /mnt/b8311c72-fe75-3832-98fc-975445028a12/01ab129f-aaf6-4b1a-8e2a-093bee0b811c.raw
> /dev/STOREY1/1e1ae58c-ca3f-4d4a-b0fd-92d18d07fb7c
> {quote}
> Since the original format was actually QCOW2 (from the previous step) this
> lead to data loss once the volume is expunged from the previous pool and from
> the nfs storage!
> Migration worked before we migrated to 4.2.1.
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
