Solution to abnormal startup of vip in RAC under SOLARIS

When the cluster is started, the first node rac1 and vip cannot be started, and other services dependent on it cannot be started. Start rac2 on the second node.

When the cluster is started, the first node rac1 and vip cannot be started, and other services dependent on it cannot be started. Start rac2 on the second node.

The task is very simple. Copy the virtual machines in Beijing to a machine on the site, let them run, and simulate the real environment for users to perform exercises. In this case, if the production database fails, O & M personnel can switch quickly. On SOLARIS, RAC is connected to a single-host replicuard, and a virtual storage is added. A total of four virtual machines are supported. It was instructor Zeng who handed it over to me. I ran it on my computer before I came to Jinan. No problem. I thought it was a very simple task to set up the environment today. As a result, I went back to the hotel only after working overtime. So depressing.

The machines used on site are an HP Workstation. First of all, when encountering operating system problems, I always feel that the XP installation is not reliable. It takes more than 20 minutes to decompress the package at home. It takes more than two hours to complete the package. Therefore, reinstall the system, install win server 2008 r2, add the memory to 8 GB, and restore the belt partition to four single disks. Each disk is placed in one virtual machine. I/O crisis eased a little. After decompression, half a day is over... start openfiler first, and you need to reconfigure the network. However, after changing the host, eth0 becomes eth1, which is really powerful. Then, start RAC1. After a long process, the related services are started in a very calm manner, but the VIP cannot survive. GSD, LSNR, and DB1.SRV are also suspended. After starting RAC2, the VIP of RAC1 finally got up. This is good. Services of RAC1 and RAC2 take turns on vacation, one node restarts the system due to the timeout of interconnection, and no useful alert information is found. The symptoms seem to be a conflict between the VIP addresses of the two nodes, which may be related to the virtual Nic, which is quite strange.

At last, we re-executed vipca and restarted the CRS service, which seems to be normal. Check again. Don't give me any more messy questions.

View next page