Difference between revisions of "Deploy Procedure in Kadeploy 3 and OAR"
From Supercomputación y Cálculo Científico UIS
| Line 8: | Line 8: | ||
It's necessary to check the exact name of the environment to deploy. The command kaenv3 -l allows to see all the environments available for the current user. | It's necessary to check the exact name of the environment to deploy. The command kaenv3 -l allows to see all the environments available for the current user. | ||
{{Command|<nowiki>kaenv3 -l</nowiki>}} | {{Command|<nowiki>kaenv3 -l</nowiki>}} | ||
| − | |||
{{Command|<nowiki>oarsub -I -t deploy</nowiki>}} | {{Command|<nowiki>oarsub -I -t deploy</nowiki>}} | ||
| − | + | In this example, the base image selected for deployment is 'squeeze-x32-base'. | |
{{Command|<nowiki>kadeploy3 -e squeeze-x32-base -f $OAR_NODEFILE</nowiki>}} | {{Command|<nowiki>kadeploy3 -e squeeze-x32-base -f $OAR_NODEFILE</nowiki>}} | ||
| − | + | Then you can access the deployed node via ssh: | |
| − | {{Command|<nowiki>ssh root@ | + | {{Command|<nowiki>ssh root@nodoXX</nowiki>}} |
'''Saving the modified image''' | '''Saving the modified image''' | ||
| Line 21: | Line 20: | ||
{{Command|<nowiki>rm /etc/udev/rules.d/70-persistent-net.rules</nowiki>}} | {{Command|<nowiki>rm /etc/udev/rules.d/70-persistent-net.rules</nowiki>}} | ||
{{Command|<nowiki>mount -o bind / /mnt</nowiki>}} | {{Command|<nowiki>mount -o bind / /mnt</nowiki>}} | ||
| − | After exiting the machine, on the front-end execute the following commands: | + | After exiting the machine, on the front-end execute the following commands, where xx is a numeric string between 01 and 16 (nodes availables at the moment) and archivoImagen.tgz is the filename of the image file: |
{{Command|<nowiki>ssh root@guanexx "cd /mnt; tar --posix --numeric-owner --one-file-system -zcf - *" > archivoImagen.tgz</nowiki>}} | {{Command|<nowiki>ssh root@guanexx "cd /mnt; tar --posix --numeric-owner --one-file-system -zcf - *" > archivoImagen.tgz</nowiki>}} | ||
| − | Create the following file in an accessible location: | + | Create the following file in an accessible location, taking into account the filename used and other similar parameters: |
{{File|ambiente.dsc|<pre><nowiki> | {{File|ambiente.dsc|<pre><nowiki> | ||
name : ambiente | name : ambiente | ||
| Line 29: | Line 28: | ||
version : 1 | version : 1 | ||
author : usuario | author : usuario | ||
| − | tarball : /home/usuario/ | + | tarball : /home/usuario/archivoImagen.tgz|tgz |
kernel : /boot/vmlinuz-2.6.26-2-amd64 | kernel : /boot/vmlinuz-2.6.26-2-amd64 | ||
initrd : /boot/initrd.img-2.6.26-2-amd64 | initrd : /boot/initrd.img-2.6.26-2-amd64 | ||
| Line 38: | Line 37: | ||
environment_kind : linux | environment_kind : linux | ||
</nowiki></pre>}} | </nowiki></pre>}} | ||
| + | In this case the image is going to be private, which means it can only be accessed and deployed by the user that created it. If you want it to be public, an administrator must do it, and must set the visibility parameter to public. Also, it is advisable to copy it to /home/images | ||
Revision as of 16:06, 11 October 2014
NOTE:
There are two kinds of environments, private and public. Private environments are only usable for the user that created them. Public environments are available for all users but they can only be created by root. It's important to note that difference and check whether an enviroment should be public or private when creating new ones. Files for public environments should be created in public directories, even if those aren't available via permissions to all users. This is a style norm.
It's important to note that the user that performs the operations related to deployments affects the
Deploying a base image
It's necessary to check the exact name of the environment to deploy. The command kaenv3 -l allows to see all the environments available for the current user.
kaenv3 -l
oarsub -I -t deploy
In this example, the base image selected for deployment is 'squeeze-x32-base'.
kadeploy3 -e squeeze-x32-base -f $OAR_NODEFILE
Then you can access the deployed node via ssh:
ssh root@nodoXX
Saving the modified image
In the machine execute these commands
rm /etc/udev/rules.d/70-persistent-net.rules
mount -o bind / /mnt
After exiting the machine, on the front-end execute the following commands, where xx is a numeric string between 01 and 16 (nodes availables at the moment) and archivoImagen.tgz is the filename of the image file:
ssh root@guanexx "cd /mnt; tar --posix --numeric-owner --one-file-system -zcf - *" > archivoImagen.tgz
Create the following file in an accessible location, taking into account the filename used and other similar parameters:
File: ambiente.dsc
name : ambiente version : 1 author : usuario tarball : /home/usuario/archivoImagen.tgz|tgz kernel : /boot/vmlinuz-2.6.26-2-amd64 initrd : /boot/initrd.img-2.6.26-2-amd64 fdisktype : 83 filesystem : ext3 visibility : private demolishing_env : 0 environment_kind : linux
In this case the image is going to be private, which means it can only be accessed and deployed by the user that created it. If you want it to be public, an administrator must do it, and must set the visibility parameter to public. Also, it is advisable to copy it to /home/images
