MicroVMs deploy

Community Support General
1

I’ve created separated datastores on the firecracker host, and install there kernel and image from docker hub.
But when im trying to deploy microvm, i got the next error:

Tue Feb 22 00:08:56 2022 [Z0][VM][I]: New state is ACTIVE
Tue Feb 22 00:08:56 2022 [Z0][VM][I]: New LCM state is PROLOG
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]: Command execution failed (exit code: 2): /var/lib/one/remotes/tm/ssh/clone vonecloud.web-bee.loc:/var/lib/one//datastores/106/94ebea217153d78700e1693ecc52ec2a fc-host.web-bee.loc:/var/lib/one//datastores/107/31/disk.0 31 106
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]: clone: Cloning /var/lib/one//datastores/106/94ebea217153d78700e1693ecc52ec2a in /var/lib/one//datastores/107/31/disk.0
Tue Feb 22 00:08:59 2022 [Z0][TrM][E]: clone: Command " set -e -o pipefail
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]:
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]: if [ -d "/var/lib/one//datastores/106/94ebea217153d78700e1693ecc52ec2a.snap" ]; then
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]: SRC_SNAP="94ebea217153d78700e1693ecc52ec2a.snap"
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]: fi
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]:
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]: tar -C /var/lib/one//datastores/106 --transform="flags=r;s|94ebea217153d78700e1693ecc52ec2a|disk.0|" -cSf - 94ebea217153d78700e1693ecc52ec2a $SRC_SNAP | ssh fc-host.web-bee.loc "tar -xSf - -C /var/lib/one//datastores/107/31"" failed: Warning: Permanently added 'vonecloud.web-bee.loc,192.168.15.95' (ECDSA) to the list of known hosts.
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]: tar: /var/lib/one//datastores/106: Cannot open: No such file or directory
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]: tar: Error is not recoverable: exiting now
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]: tar: This does not look like a tar archive
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]: tar: Exiting with failure status due to previous errors
Tue Feb 22 00:08:59 2022 [Z0][TrM][I]: Error copying /var/lib/one//datastores/106/94ebea217153d78700e1693ecc52ec2a to fc-host.web-bee.loc:/var/lib/one//datastores/107/31/disk.0
Tue Feb 22 00:08:59 2022 [Z0][TrM][E]: Error executing image transfer script: INFO: clone: Cloning /var/lib/one//datastores/106/94ebea217153d78700e1693ecc52ec2a in /var/lib/one//datastores/107/31/disk.0 ERROR: clone: Command " set -e -o pipefail if [ -d "/var/lib/one//datastores/106/94ebea217153d78700e1693ecc52ec2a.snap" ]; then SRC_SNAP="94ebea217153d78700e1693ecc52ec2a.snap" fi tar -C /var/lib/one//datastores/106 --transform="flags=r;s|94ebea217153d78700e1693ecc52ec2a|disk.0|" -cSf - 94ebea217153d78700e1693ecc52ec2a $SRC_SNAP | ssh fc-host.web-bee.loc "tar -xSf - -C /var/lib/one//datastores/107/31"" failed: Warning: Permanently added 'vonecloud.web-bee.loc,192.168.15.95' (ECDSA) to the list of known hosts. tar: /var/lib/one//datastores/106: Cannot open: No such file or directory tar: Error is not recoverable: exiting now tar: This does not look like a tar archive tar: Exiting with failure status due to previous errors Error copying /var/lib/one//datastores/106/94ebea217153d78700e1693ecc52ec2a to fc-host.web-bee.loc:/var/lib/one//datastores/107/31/disk.0
Tue Feb 22 00:08:59 2022 [Z0][VM][I]: New LCM state is PROLOG_FAILURE

Image is placed on expected path on firecracker host:

[root@fc-host ~]# ls -la /var/lib/one//datastores/106/94ebea217153d78700e1693ecc52ec2a
-rw-rw-r--. 1 oneadmin oneadmin 2147483648 Feb 22 03:07 /var/lib/one//datastores/106/94ebea217153d78700e1693ecc52ec2a

I believe that the error was, cause as a source used frontend address, but image is on firecracker host.
So, when i moved datastore 106 to the frontend node, problem solves, i believe that as a source param to tm/ssh/clone script shoud be hostname of a host with datastore instead of frontend hostname

2

HI @Jehe,

Images DS are supposed to be available at the frontend node (unless BRIDGE_LIST is configured). Then depending on the TM_MAD driver you are using images will be accessed/transferred to the hypervisor in the corresponding way. It seems that you’re using SSH driver so images will be copied via SSH from the frontend node to the hypervisor.

Note that doing what you proposed will require to replicate images DS in every HV node, instead of having images on just one location (frontend).

3

BRIDGE_LIST should have frontend node ?

Just, i’m not include frontend node into BRIDGE_LIST of datastore

4

BRIDGE_LIST must contain the address of a node with access to the storage. If the frontend have access to it you can avoid it. That’s why this works:

So, when i moved datastore 106 to the frontend node, problem solves