Famous collection-eagle-a bit of experience on Troubleshooting

eagle-a bit of experience with troubleshooting
troubleshooting has been working in the company, and there are often some new problems, but there are also some difficult problems. Let's talk about your own experience in performing troubleshooting. I hope it will be helpful to you.

first, we must have a solid Oracle Foundation and have a deep understanding of some basic concepts. I recommend that you read books in the first place based on English documents, supplemented by some original articles . Maybe you have a poor foundation in English and it will be difficult to read at the beginning. However, after a long time, you will realize how wise you were to insist on reading English documents. DBAs around me are reading English books. Now I can't read Chinese translation books, but I don't know what he is talking about. On the contrary, I can see English fluently. If you want to be a good dBA, you must read books in English. Many people also recommend a lot of books for you. Let me talk about my recommendation list here. The first is concept, Administrator's Guide, and performance tuning in the official documents. The official documents can be downloaded from otn.oracle.com. If you are afraid of a huge number of British books, you can read them from the OCP books first, and then read the three books. Many people know that these three books are good, but they really don't know much about them, and they really don't understand them much. If you understand it, congratulations, you have a good level. I recommend a few other good books.
the first is Jonathan Lewis's two books, practical Oracle 8i (I have never seen an electronic version of this book. If anyone has any, please let us know) and cost based Oracle fundamentals. Jonathan Lewis has always been the first norm of the Oracle Session I think. Both of his books are classic. He is looking forward to two other CBO books.
Steve Adams's masterpiece Oracle8i. Internal. services. For. Waits. latches. Locks seems to be Steve Adams's only book. This book is very thin, but concise. It is very beneficial to read more.
K Gopalakrishnan's oracle. wait. interface. a. practical. guide. to. performance. diagnostics. and. tuning
This book can also be downloaded from itpub. It is a very practical book that provides a good explanation of some common Oracle wait events. Required for performing troubleshooting.
do not blindly pursue progress when you are studying.
at this stage, you need to read more books and think more.

Step 2: With solid basic skills, we must strengthen practice. At this time, we need to do a lot of things to strengthen basic knowledge. If you don't need something you 've seen for a long time, it's easy to forget it. You can understand it in depth only by constantly using it in practice. This stage requires a certain degree of work environment. If your current work environment is not strong enough and the workload is not large enough, you are advised to consider another job. As long as you play well in the first stage, you can't find a suitable environment. In fact, some interview questions of major companies are at the bottom of basic skills. Do not blindly do things in the process of doing things. Do things with thinking. Do not underestimate some simple things, such as add column, create/drop table, insert/delete/update, etc. In fact, more than 90% of what DBAs usually do are simple things, however, it is difficult to do a lot of things without making mistakes. You need to think about the impact of doing this, what problems may occur, how to solve the problem, and how to roll back it. You need to think clearly before you start, instead of waiting for the problem to begin consideration.
Some people make some very low-level mistakes on the Internet. Do not underestimate these mistakes, or use the excuse "I don't understand, I just don't know, I'm not careful" to help myself. The subtitle of my website is "We are what we repeatedly do. excellence, then, is not an act, but a habit. "I first saw this sentence on the company's outage plan, the one from Aristotle. Famous saying . People are made up of habits. The vast majority of things a person does a day are repeating his or her habits, which are also the most difficult to develop. However, DBAs must develop a careful habit, which is more important than learning more about it. In addition, for some knowledge, I think it is of little significance to focus too much on some internal things that have nothing to do with practical applications. Some people do not know how to add online after online create/rebuild index on the production system, but what they say to delay block cleanout is true. I am not saying that internal research is not good, but it is necessary to distinguish the primary and secondary, and focus on practicality. In this regard, I admire Jonathan Lewis. He has a strong knowledge of internal (mentioned in Steve Adams's book), but his articles are mainly practical, he rarely publishes practical things.
In addition to the development of habits, we should also pay attention to the collection of some details at this stage. Oracle management has a lot of scattered knowledge, and most of these scattered knowledge is learned in a large number of intensive practices.
At this stage, we need to do more and make fewer mistakes.

The above two points are actually preparations. The following describes the troubleshooting experience.

1. Do not ignore any details. When a problem occurs, there may be multiple phenomena, such as many active sessions, many enqueue, and many latch free. If your explanation cannot explain every phenomenon or cover every small detail, it is very likely that your explanation is wrong and you need to overturn it and rereason. Anyone who has read conan or Jin Tianyi's animation knows that if there is a problem that doesn't work out why, they won't be able to make the final conclusion.

2. Be full of imagination. For simple problems, you may be able to see the cause at a glance, but it is difficult for some complicated problems. Many phenomena are mixed together, so it is difficult to identify which is the cause and which is the result. At this time, we need to think boldly and combine all the phenomena to give a reasonable explanation. Be good at challenging your existing knowledge. It may be because of database version changes or Oracle bugs that are no longer appropriate for you to solve some problems. At this time, you cannot stick to your own knowledge. You must be good at overthrowing yourself and go up in a spiral.

3. Strict reasoning. It is about evidence that we cannot solve the problem simply by imagination. If your explanation is correct, find evidence to support your explanation. You can use the mining of other phenomena or subsequent experiments to prove your interpretation. For some bugs, you must find the relevant documents or open tar to get Oracle confirmation. If you do not encounter any strange problems, it is an Oracle bug. There are a lot of Oracle bugs, but what you may encounter is not a bug, it may be a lack of knowledge.

4. Think deeply. Some problems are hard to imagine at the beginning, but when you go deep into one layer, you feel that all the original phenomena can be perfectly explained. I am deeply touched by this point during the troubleshoot process. You can give an example later.

5. Focus on the solution. After finding the root problem, the solution is easier. The main reason is that the basic skills must be solid and the key points must be grasped.

Avoid two points:

Do not be confused
The basic concepts must be clear, such as pctfree, pctused, LMT, and DMT ." Cache buffer chain "and" buffer busy wait "are two completely different wait events, specifying a large rollback segment for select does not help the ora-01555, and so on.

Do not go to the Internet to find an example of someone else. You must thoroughly understand the cause and effect of a problem to make a correct judgment. Other people's examples can only be used for reference.

Finally, make a small advertisement: If you feel that your basic skills are solid but you lack a suitable environment, please join us. Our system is busy, with many things and many challenges.
Send me an email: xfan (AT) ebay.com.