| havelq 回复于:2003-04-08 14:50:40 |
| 如何从swap中分出30M空间作为新分区来封装rootdg, 帮!! |
| havelq 回复于:2003-04-08 20:05:35 |
| 有人帮吗 |
| nosorryforlove 回复于:2003-04-08 21:49:44 |
| 我们可以从根盘(假设它是c0t0d0s2)“偷”出一点空间来,再做出一个slice,比如是c0t0d0s7,最好从swap的尾部分出,假设swap slice从柱面1000到3000,那么你可以把2900到3000分给slice7,具体要根据你的硬盘情况,但要保证slice7至少有10M的大小。用format命令完成这一步。
然后用以下步骤重新做一个rootdg: 1.首先建立一个文件: #touch /etc/vx/reconfig.d/state.d/install-db 2. reboot system 3. input following commands: #vxconfigd -m disable #vxdctl init #vxdg init rootdg #vxdctl add disk c0t0d0s7 type=simple #vxdisk -f init c0t0d0s7 type=simple #vxdg -g rootdg adddisk c0t0d0s7 #vxdctl enable 4. 删除文件install-db # rm /etc/vx/reconfig.d/state.d/install-db 5.reboot system again 你试试吧 |
| havelq 回复于:2003-04-09 09:08:09 |
| 谢了!!!
我正在学veritas,它有针对sun hp_ux aix。。。 好多,我现在只有sun环境,在其它系统上的做法相差大吗?? |
| nosorryforlove 回复于:2003-04-09 09:27:14 |
| Veritas的产品分2大部分,一大块叫是online的,象VxVM(也就是Volume Manager)和VCS(Veritas Cluster),另一大块是offline的,象Backup。
但它们都是多平台的产品,差别不是很大,象VxVM,学习主要是概念的建立,只要你的概念清晰,具体的命令是很简单的。 |
| havelq 回复于:2003-04-09 09:52:59 |
| 公司准备要我考veritas的认证,我现在有产品内的pdf文档,这些资料够用吗? |
| nosorryforlove 回复于:2003-04-09 10:02:09 |
| 认证应该有相应的教材,如果你做Veritas产品支持的时间不长,但又必须过那个认证,还是有针对性地准备比较好,这和在学校应付考试一个道理。但如果你是想掌握技术,那还是边做边学习docs和manual是踏实的途径。 |
| havelq 回复于:2003-04-09 10:09:42 |
| 明白了,谢!!! |
| havelq 回复于:2003-04-09 13:50:12 |
| 再问个问题
在rootdg之外,建一个datadg之后,用这个datadg就能将其它disk全包括了,如果这些空间不全分给datadg,而是又建了data01dg、data02dg,此种情况,多个dg共存有什么好处与坏处吗? |
| nosorryforlove 回复于:2003-04-09 15:15:28 |
| DiskGroup主要作用是方便管理:
第一,从用户的角度看,可以根据应用把磁盘分到不同的DiskGroup中去,例如app1dg,app2dg等等; 第二,从VxVM本身的功能看,在双机或多机的环境下,它还可以以DiskGroup为单位进行“切换”;在许多Cluster软件就利用了这点,当一台主机发生故障时,Cluster首先会停应用,然后停Volume,最后是交出对磁盘的所有权,在VxVM中就是将DiskGroup “deport” 出去;然后Cluster会在另台主机上首先做 ”import“ DiskGroup的动作,然后起Volume,最后再起应用; 第三,上面所说的deport和import的功能也可以从命令行由用户来手工调用,也就是说用户可以根据需要来手动地切换diskGroup;例如,你有2台机分别跑Oracle应用和Sybase应用,分别有2个diskgroup存放2个应用的数据,分别叫ora-dg和syb-dg。当你需要在一台机进行系统维护工作而停机的时候,你可以将应用手动切换到另一台机: 在A机: #vxdg deport ora-dg 在B机: #vxdg import ora-dg 大致如此吧。 |
| nosorryforlove 回复于:2003-04-09 15:15:45 |
| DiskGroup主要作用是方便管理:
第一,从用户的角度看,可以根据应用把磁盘分到不同的DiskGroup中去,例如app1dg,app2dg等等; 第二,从VxVM本身的功能看,在双机或多机的环境下,它还可以以DiskGroup为单位进行“切换”;在许多Cluster软件就利用了这点,当一台主机发生故障时,Cluster首先会停应用,然后停Volume,最后是交出对磁盘的所有权,在VxVM中就是将DiskGroup “deport” 出去;然后Cluster会在另台主机上首先做 ”import“ DiskGroup的动作,然后起Volume,最后再起应用; 第三,上面所说的deport和import的功能也可以从命令行由用户来手工调用,也就是说用户可以根据需要来手动地切换diskGroup;例如,你有2台机分别跑Oracle应用和Sybase应用,分别有2个diskgroup存放2个应用的数据,分别叫ora-dg和syb-dg。当你需要在一台机进行系统维护工作而停机的时候,你可以将应用手动切换到另一台机: 在A机: #vxdg deport ora-dg 在B机: #vxdg import ora-dg 大致如此吧。 |
| slm072 回复于:2003-04-09 15:35:37 |
| 不错 |
| havelq 回复于:2003-04-09 15:41:42 |
| 遇到高手了,多问几个问题,哈哈
在vvr中有同步与异步两种,我看文档中有这样一句对异步的话: Additionally, the asynchronous component is responsible for writing the request to the Primary data volume. This operation is performed asynchronously to avoid adding the penalty of a second full disk write to the overall write latency. Because the log write, but not the data write, is performed synchronously, the SRL becomes the final arbiter as to the correct contents of the data volume in the case of a system failure. 没太理解,帮一下啊! |
| havelq 回复于:2003-04-09 15:54:50 |
| 啊,好象是明白了,它是说在对主卷发写请求时是异步的,而对srl是同步的,srl是决定系统失败的仲才者
对不对 |
| nosorryforlove 回复于:2003-04-09 16:28:57 |
| 首先你要理解VVR的同步和异步的含义,以及各自的优缺点。
在异步模式下,只有保证Log是同步的,才能保证最终的数据卷的完整性。 而对数据卷的操作必须是异步的,这个异步是相对于本机同步LOG写而言的。 否则会增大异机backup volume和本机primary volume之间的latency。 这样理解对吗? |
| nosorryforlove 回复于:2003-04-09 16:40:48 |
| 以下内容摘自Veritas网站关于VVR的FAQ:
[b:b533ab7a09]The synchronous mode[/b:b533ab7a09] also known as the hard synchronous mode of replication, ensures that the Secondary host has acknowledged an update before completing the update at the Primary. In case of a problem such as a .network failure, it ensures that the update fails at the Primary itself. The synchronous mode of replication is most effective in the application environments: - that have lower update rates but require all the hosts to always reflect the same data. - where delay in updates between the Primary and Secondary host is not aclearcase/" target="_blank" >cceptable. [u:b533ab7a09]Advantage: [/u:b533ab7a09] In the event of a disaster at the Primary host, data can be recovered from the surviving Secondary host without any loss, because the Primary and the Secondary host contain the same data. [u:b533ab7a09]Disadvantage: [/u:b533ab7a09] The response time experienced by the writing application is affected because the application has to wait for an acknowledgment from the Secondary before it can complete an update. The following suggestions will help to work around the disadvantages to a certain extent: - Add network bandwidth to reduce the degradation in update response time experienced by the application. - Reduce the network round-trip time between each Primary and Secondary pair by using faster network technologies. [b:b533ab7a09]The asynchronous mode[/b:b533ab7a09] the application updates are immediately reflected at the Primary, but are sent to the Secondary with some delay. The updates are stored in the Replicator Log until they are sent to the Secondary. If the writing application experiences temporary increase in update rate, this delay may increase. If a disaster strikes during a period of peak update activity, it is possible that the most recent updates at the Primary host are not reflected in the data at the Secondary host. This is because of the lag between the Primary and Secondary data states, also called latency. To prevent this, you can configure the latency such that the data inconsistency will be acceptable in the event of a disaster. Asynchronous replication ensures that the lag never exceeds this configured maximum. [u:b533ab7a09]Advantages: [/u:b533ab7a09] - The writing application in Asynchronous replication does not suffer from the response time degradation caused by each update. Therefore there is no network round-trip overhead. - At all times, the Replicator Log is being drained as fast as the available network bandwidth allows. During periods when the update rate is less than the available network bandwidth, the Replicator Log drains faster than it grows. This allows the Secondary data state to catch up with that on the Primary. - Assures that all completed updates to the Primary volumes will be made on the Secondary data volumes, even though it may be with some delay. This is true even in case of failures in communication or system crashes on any of the participating hosts. - Asynchronous replication can easily handle the temporary network or the Secondary host failure because of its ability to queue updates persistently, and hold them at the Primary for later transmission. [u:b533ab7a09]Disadvantages: [/u:b533ab7a09] - The improvement in response time is at the cost of the data at the Secondary host lagging behind the data on the Primary host, during peak update times. - The data at a Secondary host may not have the latest updates when a failover occurs. |
| havelq 回复于:2003-04-09 21:56:32 |
| 我要从分区中挤出空间:
#format 选 0 #partition 选7 然后我改缩小块区间 然后保存后, 退出format , 可再进入format后,verify空间没变化,怎么回事啊, !!!! |
| nosorryforlove 回复于:2003-04-09 23:24:30 |
| 兄弟,应该很简单呀!
先把一个slice缩小了吗?然后把让出的给另一个就行了啊! 退出partition之前你label了吗? label后有没有再P一下检查啊? format命令很危险,小心点! |
| sunmarmot 回复于:2003-04-10 01:34:43 |
| up~~~偶这两天也在搞这个,坚决不能让它沉下去 |
| havelq 回复于:2003-04-10 08:43:32 |
| ,又有一事相求:
我在机器上的第一块盘上用一个分区做rootdg,里面不装其它的东西。第二块盘上做一个应用dg,用于存应用程序(如oracle的应用程序),然后在a1000上做数据dg,用于装数据和log, 数据和log在不同的卷上, 你看这样做是不是合理的。 |
| nosorryforlove 回复于:2003-04-10 12:24:02 |
| 你这样做应该是没问题的。
不过我碰到的许多情况是: 如果机内有2块硬盘,那就把它们都放入rootdg,然后做mirror。 至于A1000,我觉得没有必要交给VxVM管理了,因为它已经是硬件RAID了。 除非你有其他考虑。 |
| havelq 回复于:2003-04-10 13:14:59 |
| 请问
我做vvr要不要求2台机器和阵列的空间分配都一样,包括各自的dg, rootdg的卷及卷大小,其下的disk 谢 |
| nosorryforlove 回复于:2003-04-10 14:30:38 |
| [b:811c167b9d]这些可以从安装文档中看到吧[/b:811c167b9d],建议你先看看相关文档吧。 |
| havelq 回复于:2003-04-11 09:43:40 |
| vvr是不是只能实现复制,为恢复做准备,而要想做到应用的热切换要依靠vcs来实现了?
谢 |
| nosorryforlove 回复于:2003-04-11 12:06:57 |
| yes |
| bear 回复于:2003-04-11 12:40:20 |
| 不错,补充一下,调整系统盘的分区,用光盘引导起来做。否则MOUNTED的分区的硬盘是无法LABEL的。 |
| sunmarmot 回复于:2003-04-11 13:30:04 |
| 谢谢bear的提醒~~~~~~~~~ |
| havelq 回复于:2003-04-11 13:57:03 |
| 在vvr中mirror功能是什么啊, |
| zzsir 回复于:2003-04-13 21:17:03 |
| ? |
| havelq 回复于:2003-04-14 13:50:45 |
| 在做vvr复制时,文档要求先做DCM,没太理解DCM和如何做DCM
谢 |
| havelq 回复于:2003-04-14 14:12:38 |
| 我现在把VVR的配置都看完了,可是在命令上(无论是cli还是vvr-as方式)都没看出哪个是启用同步或异步的方式复制。
以带检查点的全同步方式开启复制。 # vradmin -g hrdg -c chkpt_presync startrep hr_rvg london 用自动同步方式开始复制 # vradmin -g hrdg -a startrep hr_rvg london 用-f选项开启复制。 # vradmin -g hrdg -f startrep hr_rvg London 在哪里做呢 |
| zhaogzh 回复于:2003-04-18 17:17:37 |
| 啥,我成了老土了,我怎么不懂你们的意思啊?哎,两年没用SUN就成文盲了。 |