norcams/iaas

IaaS project documentation

norcams/iaas

This repository is our team handbook and the documents here are built as HTML and published to

https://iaas.readthedocs.org

iaas's Issues

Naming when launching multiple instances

If you set the number of instances higher than 1 a suffix is automatically added to each instances' name (-1, -2, etc). However, it doesn't take into account if you another have an instance with that exact number already.

Possible solutions

• Skip any present names until a "free" number is found.
• Add a field in the instance launcher so the user can select which instance number to start with.

Newly built instances not responding

Hi,

I have been trying to build a new instance for a couple of hours, they seem to be spawn and running but I cannot access them with ssh.

I need it for a demo tomorrow and would appreciate if that worked quickly now, since I have to reinstall a lot of stuff.

Can you please have a look?

Thanks,

Jean

Automatically test file system sematics

You might find http://people.skolelinux.org/pere/blog/Testing_if_a_file_system_can_be_used_for_home_directories___.html and http://github.com/gebi/fs-test to test the Ceph file system sematics.

ansible inventory via `terraform.py` picks up instances from unrelated sibling directories

Hi, the terraform.py script linked from https://docs.nrec.no/terraform-part4.html#ansible-inventory-from-terraform-state contains the following:

   default_root = os.environ.get('TERRAFORM_STATE_ROOT',
                                  os.path.abspath(os.path.join(os.path.dirname(__file__),
                                                               '..', '..', )))

When following the instructions for linking it as inventory/hosts of the terraform working directory, ansible all -i inventory --list-hosts will also pick up hosts from the .tfstate files of sibling directories of the current directory (since default_root will be the parent directory of the current directory).

As a workaround, one can add the line export TERRAFORM_STATE_ROOT=$(pwd) to keystone_rc.sh.

Email warning on instance terminating?

Is this a good or bad idea?

H1 - title element in docs is covered by navbar

The title in the documentation is covered by the top nav-element.

This can be fixed by assigning position:relative to the navbar - element (but loose the fixed menu on top). If you want to keep the fixed menu on top, the padding-top on the body-element must match the height of the navbar for different screen widths, and the navbar should probably have a fixed height for each screen-width interval.

Reference header image source

You should document that the header image for iaas.uhsky.no is taken from wikipedia (CC-by-SA)

https://no.wikipedia.org/wiki/Tyssedal_kraftanlegg#/media/File:Kontrollrom_Tyssedal.jpg

https://github.com/norcams/iaas/blob/master/site/public/img/header.jpg

Need for floating IP addresses

Just logging case for documenting requirements using IaaS-instances as compute nodes in clusters .

During testing of Openshift origin installation in IaaS it became evident that IP addresses only work on the instance it is assigend to in the Openstack API. This is apparently by design. However, if IaaS is going to be used for building clusters, there will often be a need for IP addresses that the cluster can move around between the nodes by itself.

At the moment this is not possible, at least not in a documented manner.

Some references related to storage fail tolerance

As discussed, here is a list of references.

:login; - Redundancy Does Not Imply Fault Tolerance
Analysis of Distributed Storage Reactions to Single Errors and Corruptions

Aishwarya Ganesan et al
https://www.usenix.org/publications/login/summer2017/ganesan

ZDNet - Why RAID 5 stops working in 2009

Robin Harris
http://www.zdnet.com/article/why-raid-5-stops-working-in-2009/

ZDNet - Why RAID 6 stops working in 2019

Robin Harris
http://www.zdnet.com/article/why-raid-6-stops-working-in-2019/

USENIX FAST'07 - Failure Trends in a Large Disk Drive Population

Eduardo Pinheiro, Wolf-Dietrich Weber and Luiz André Barroso
http://research.google.com/archive/disk_failures.pdf

Data Integrity
Finding Truth in a World of Guesses and Lies

Doug Hughes
https://www.usenix.org/system/files/login/articles/hughes12-04.pdf

See also "Producing Open Source Software", a nice book about how to
organize a project. Available from http://producingoss.com/.

Cannot create snapshot

I have a running instance and wish to create a snapshot from it with "Create Snapshot". It runs for a while; my image starts to be created and then with no message it disappears and nothing is created.

It worked with instances I have created before. So I don't know why it does not work anymore.

I installed software and libraries we are using at the Department of Geosciences so I really would like to be able to save my image before my instance stops running.

Can you help?

Thanks,

Anne.

'Time since created' mismatch for older instances

The dashboard views Overview and Instances both lists your instances and their time since created.

Strangely, the times for my older instances are different in the two views. It seems to me the times in Instances are correct and the ones in Overview are way off.

#	Time - Overview	Time - Instances
A	2 weeks, 2 days	1 year, 10 months
B	1 day, 2 hours	1 day, 2 hours
C	2 weeks, 2 days	1 year, 1 month
D	3 months	3 months
E	2 weeks, 2 days	1 year, 3 months

Requirements for openstack CLI to work on RHEL7

Attempted to use openstack client on RHEL7 by:

• pip install python-openstackclient
• Download «Openstack RC File"
• Source it (and provide password)

Result:

jb@watershed ~ $ openstack 
(openstack) image list 
Expecting to find domain in user - the server could not comply with the request since it is either malformed or otherwise incorrect. The client is assumed to be in error. (HTTP 400) (Request-ID: req-b05965cc-1d1c-4685-aa77-aa13ee38457a)
(openstack)