[
https://issues.apache.org/jira/browse/CLOUDSTACK-4746?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
manasaveloori closed CLOUDSTACK-4746.
-------------------------------------
steps:
1. Deployed CS with 1 cluster having 2 hosts of same configuration.
2. Deployed some VMs using HA enabled service offering.
3. Set the GS cluster.cpu.allocated.capacity.disablethreshold=0.30 and
capacity.skipcounting.hours=3600sec and recreate.systemvm.enabled=true
http://10.147.38.243:8096/client/api?command=listCapacity&clusterid=33022899-f0db-4647-8634-7b0eccb91097
Shows cluster usedcapacity as "28.71"
mysql> select * from op_host_capacity where capacity_type=1;
+----+---------+----------------+--------+------------+---------------+-------------------+----------------+---------------+----------------+---------------------+---------------------+
| id | host_id | data_center_id | pod_id | cluster_id | used_capacity |
reserved_capacity | total_capacity | capacity_type | capacity_state |
update_time | created |
+----+---------+----------------+--------+------------+---------------+-------------------+----------------+---------------+----------------+---------------------+---------------------+
| 2 | 1 | 1 | 1 | 1 | 4500 |
0 | 9580 | 1 | Enabled | 2013-11-14
11:43:25 | 2013-11-13 16:48:45 |
| 9 | 4 | 1 | 1 | 1 | 0 |
5500 | 9576 | 1 | Enabled | 2013-11-14
11:43:27 | 2013-11-13 16:58:23 |
+----+---------+----------------+--------+------------+---------------+-------------------+----------------+---------------+----------------+---------------------+---------------------+
4. Now trigger HA.
All the VMs got migrated.
select * from op_host_capacity where capacity_type=1;
+----+---------+----------------+--------+------------+---------------+-------------------+----------------+---------------+----------------+---------------------+---------------------+
| id | host_id | data_center_id | pod_id | cluster_id | used_capacity |
reserved_capacity | total_capacity | capacity_type | capacity_state |
update_time | created |
+----+---------+----------------+--------+------------+---------------+-------------------+----------------+---------------+----------------+---------------------+---------------------+
| 2 | 1 | 1 | 1 | 1 | 5000 |
0 | 9580 | 1 | Enabled | 2013-11-14
11:44:16 | 2013-11-13 16:48:45 |
| 9 | 4 | 1 | 1 | 1 | 0 |
500 | 9576 | 1 | Enabled | 2013-11-14
11:43:48 | 2013-11-13 16:58:23 |
+----+---------+----------------+--------+------------+---------------+-------------------+----------------+---------------+----------------+---------------------+---------------------+
2 rows in set (0.00 sec).
5.set capacity.skipcounting.hours=30sec.
6.Now triggered HA.All the VMs got migrated and
mysql> select * from op_host_capacity where capacity_type=1;
+----+---------+----------------+--------+------------+---------------+-------------------+----------------+---------------+----------------+---------------------+---------------------+
| id | host_id | data_center_id | pod_id | cluster_id | used_capacity |
reserved_capacity | total_capacity | capacity_type | capacity_state |
update_time | created |
+----+---------+----------------+--------+------------+---------------+-------------------+----------------+---------------+----------------+---------------------+---------------------+
| 9 | 4 | 1 | 1 | 1 | 5500 |
0 | 9576 | 1 | Enabled | 2013-11-14
11:54:23 | 2013-11-13 16:58:23 |
| 11 | 1 | 1 | 1 | 1 | 0 |
0 | 9580 | 1 | Enabled | 2013-11-14
11:55:39 | 2013-11-14 11:55:39 |
+----+---------+----------------+--------+------------+---------------+-------------------+----------------+---------------+----------------+---------------------+---------------------+
Working fine.Closing the issue.
> Allocation capacity of a cluster during HA
> ------------------------------------------
>
> Key: CLOUDSTACK-4746
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-4746
> Project: CloudStack
> Issue Type: Bug
> Security Level: Public(Anyone can view this level - this is the
> default.)
> Affects Versions: 4.2.0
> Reporter: Nitin Mehta
> Assignee: Nitin Mehta
> Priority: Critical
> Fix For: 4.2.1
>
>
> When the host goes down. Cloudstack will stop the VM's for which HA is not
> enabled. Once cloudstack marks these VM's as stopped then the capacity of
> these VM's is moved into the reserved capacity.
> If the VM's are HA enabled then cloudstack will stop and start the VM's on
> another host. During this time the capacity of the VM's will be moved to
> reserved capacity once the VM's are stopped, But when the Vm's are being
> started cloudstack will try to calculate what will be total allocated
> capacity if the VM is started again ( even though the CPU for the VM is
> already reserved). This is a bug in cloudstack where the CPU required for the
> VM is considered twice when calculating the allocated capacity.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
