/etc/snmp/snmptrapd.conf
Predictable_vNIC_Order--可预见的虚拟网卡排序
描述:
为了解决当ovirt客户端加入一块新的vNIC时,MAC地址和PCI地址与网卡name的映射混乱,
主要的分为下面几种情况:
1. create vm from template
2. create vm from snapshot
3. import vm
当对VM添加网卡时,会有一个[name]-[MAC]的映射关系,oVirt依据[MAC]的顺序来传送设备信息给[libvirt],libvrit会分配一个[pci]地址给它。[mac]和[pci]被虚拟机的操作系统所使用以获取vNIC的[guest]名称,例如EL5/6使用[mac],EL7/fedoras使用[pci]。需要控制网卡启动顺序,目前[boot]匹配的是[mac]。
Benefit:
假设我们有两个网络,RED(机密网络)/BLUE(公开网络),部署几台用于入侵检测的VM,监控BLUE网络并发送报告至RED。创建VM,eth0-RED,eht1-BLUE。根据这个VM创建template,然后从这个VM clone出另一台VM。如果没有Predictable_vNIC_Order,这个clone出来的VM可能会eht0-BLUE,eth1-RED,然后系统将会把机密网络的信息泄漏到公开网络上。That's very bad.
1. http://www.ovirt.org/Features/PMHealthCheck --增加周期性的电源管理安全检查关于'link-down'的检测/警告的检查
对配置的PM的hosts增加一个周期性的安全检测,通过启动一个计划任务,周期性(预定义的one hour)的给PM发送一个关于状态的command,其中遇到error会raise alerts
2. http://www.ovirt.org/Features/Design/DetailedPMHealthCheck --engine代码,增加一个类PmHealtCheckManager去处理上面的那个计划任务
配置:
PMHealthCheckEnabled (boolean, false by default) - Enable/Diable the Pm Health Check scheduled job
PMHealthCheckIntervalInSec (int, default 3600) - Determines the number of seconds for scheduling the PM Healt Check operation
它主要的逻辑:
Search for all Hosts with defined and enabled power management
For each Host
If the Host has just a Primary card, send a status command to this card, In case that this failed
and Alert is generated, in case that it succeeded we check if there is an active alert for this host
and remove it.
If the Host has Primary & Secondary cards
For sequential devices, both are tested but only warning alerts are generated if one of those
cards is OK and one fails
For concurrent devices both are tested and alert is generated no matter which card fails
3. http://www.ovirt.org/Custom_Fencing
engine-config -s CustomVdsFenceType="zzz"
engine-config -s CustomFenceAgentMapping="zzz=ipmilan"
engine-config -s CustomVdsFenceOptionMapping="zzz:port=ipport"
engine-config -s CustomFencePowerWaitParam="zzz=power_wait"
4. http://www.ovirt.org/Features/Design/CommandCoordinator
CommandCoordinator提供将非存储相关操作persist/load到DB里,在3.5里面特别用于'live merge';
它用来存储SPM和NON SPM的任务,好处-当在执行'live merge'或者需要长时间的任务时engine重启了,它可以用来保存和重载任务。
设计:
1. 创建一个command_entities表
2. 提供一系列的方法用来'persist/retrieve/delete' command
3. command实体的DAO(数据访问接口)
void saveOrUpdate(CommandEntity commandEntity);
void remove(Guid commandId);
void removeAllBeforeDate(Date cutoff);
void updateExecuted(Guid id);
void updateNotified(Guid id);
void updateStatus(Guid command, Status status);
5. http://www.ovirt.org/Features/Snapshots_Overview 3.5 还在开发中
快照预览和管理存储域。
在'Storage -> Disk Snapshots Sub-Tab',可以看到你对磁盘做的快照。