| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Usenet -> c.d.o.server -> Re: Oracle Parallel Server 8.1.7
On Wed, 10 Dec 2003 17:54:11 +0300, Olesnitskiy Taras wrote:
> óÐÁÓÉÂÏ Vadim Lejnin ÚÁ ÐÏÍÏÝØ.
Taras, please post in US7ASCII code. Your post is unusable. Anyway, I have extensive OPS experience and I believe I can guess what's your problem.
My OPS experience is on HP-UX and SUN, and I don't have any jAIX experience. What I see here is the following:
DLM starts up, sets up receive buffer, prints version and attempts to communicate with the other instances. This attempt fails, probably because of the piece called "cluster manager", which is produced by the OS vendor, in this case IBM. Cluster manager can fail to join cluster for many reasons. The most frequent reasons are: incorrect setup (most often, SHMMAX too low. Those things require huge amounts of shared memory.), improper setup of the private communication cards or even hardware failure. Each of those cluster managers must have a log file which contains precisely what happens. Try typing "dlmstat" and see what you get. Is your LVM set up to use raw devices shared volumes? Are you volumes set up properly? You need a good system administrator and quickly. Last but not least, oracle may require specific patches to be installed on or jAIX box. Those patches are usually cluster manager specific or network card specific. You see, oracle opens private communication channels with other nodes in the cluster bypassing the public network layer and frequently manipulating the interface directly. In order to do that, it may need some patches. Read the fine manual about clustering on jAIX and get yourself a good system administrator.
-- None of us is as dumb as all of us. (http://www.despair.com/meetings.html)Received on Wed Dec 10 2003 - 23:02:04 CST
 |
 |