Application of O & M work order

One . explain

The first to enter the industry after training 2 In the year , Most of the companies I entered and those I met during the interview were machines 200 Less than virtual machines , O & M plus me 1-2 personal .

It's up to you to decide , Just remember what you did , Adding permissions is all about development. I'll add them here , Process coordination depends on mouth to mouth transmission , Of course, you can also use qq And wechat .

The working environment is still very important , Now there are many projects to be operated 5 individual , virtual machine 300 Upward , There are a lot of other cloud products to maintain . It is necessary to divide the work , Instead of everyone doing it when they are idle , That will cause the demander to find no one in charge , And the person in charge will change back and forth .

Then the demand comes , The following problems are found according to daily work ： 1. Developers don't know who can do this 2. Development to apply for adding permission 、 use qq And so on 3. Because everyone is responsible for one piece , All involved in the work , No one knows the overall progress 4. A certain operation and maintenance department has done some operations that others don't know 5. He is responsible for this , Now he asked for leave , I don't know what to do after taking over

Let's see how development solves the above problems ： 1. The product manager needs to raise jira 2. stay jira There is a task list on the , You can see the tasks that everyone has received , Which piece is involved 3. After the code is written, it will be merged into master The branch should be confirmed by the development team leader , Notes are also written a lot

There will also be corresponding workflows , Track every requirement from beginning to end , Later, the online test environment and even the production environment have corresponding management processes .

Two . First generation design

The first thing I thought about was inside confluence（ A note taking website ） Write down what tasks we have to do , By date , Write it down every day , Write the name after the task name .

Of course, for external needs , For example, add permissions , Solve some jenkins The configuration errors on the distribution board are all completed, and they are supplemented here .

Now think it's really stupid , The difference between recording and not recording is not big at all , Although I know who did what at what time , But it's hard to maintain , Interaction with other groups is also primitive mouth to mouth communication .

At the same time, I can't see [ Task schedule ]、[ Who is paddling ]、[ Number of tasks of a certain type ]、[ Number of failures in the current month ] wait , That is to say, only , But statistics are very difficult , Manual counting is required .

In discovery development jira after , Explored , It turns out that it really works , Can realize our needs . Through matching nail notice and reading version , Then establish a clear and perfect workflow , You can avoid spending time on the development process with maximum efficiency .

3、 ... and . Effect display

use jira establish 2 A project , The other is the external work order used to process external requirements , One is to use only internal recording tasks for internal work orders .

The reason to distinguish them is that they are mixed up in a mess , It's easy to say internally that only a few people create categories according to tasks , For example, if you buy a server, you can set up [ fund management ] Mission , To deal with the fault, we establish [ Fault handling ] Mission .

But development doesn't care , I don't care whether it is to apply for permission 、 Deal with errors 、 Application resources are all [ Fault handling ] This default question type . So it should be reduced to 2-3 Class best , Here are permissions and other types .

External tasks ： 1. Permission to apply for , Here, the permission application will be accompanied by the nailing script for timeout notification , Users should renew their permissions regularly , Of course, you can adjust the larger options , for example 6 Months . Add applicant options separately , Because the applicant may not have jira, Or outsourcing personnel

When the authority exceeds the specified period , Will give jira Both the job publisher and the administrator of send a pinning message

2. Other tasks

Internal tasks ： 1. Service management , Here is the internal task , For example, publishing 、 Build services 、 Backup data, etc . For a big task, it must be broken down into versions , It's a collection , Refine the task .

Or you can just write something k8s Optimization is a false big empty , No goal is the same as a stroke , do 2 I feel irritable and don't bother to do it , It doesn't make sense . At the same time, for important operations such as production , It is necessary to write the supporting tasks 《 Operation case 》, You didn't guess wrong. , It is the benchmarking test 《 The test case 》. Because operation and maintenance does not require fast and stable , Document operation does not cause any problems , It's much cheaper to fix than to fix problems .

2. Alarms can be selected in several categories , I will do it later here zabbix Linkage establishment jira Mission , Actually zabbix Each alarm can be processed and recorded by itself , But put it in jira Unified management .

3. Other categories , It's all my own people , You can make the problem more detailed , This is good for later use jira Filter for statistics . For example, count how many times a person has published this month , A few mouse clicks will show the results .

For example, since I created the work order , The official production and distribution covers a total of 10 Time

Four . Work order operation process

For external jobs , The default operator is the operation and maintenance team leader , When we get to him , See the nail notice , Then assign follow-up tasks , Mobilize people .

According to jira Query each person's task volume in , Until then grafana On the website , I haven't finished here yet , The following figure will be supplemented , It is equivalent to doing internal 《 Kanban 》, Visualize the task status of everyone in the market .

This will be clear at a glance through the large screen kanban , You can find out who has the least tasks this week , Or procrastinate , Originally [ELK Log analysis ] Build this big project , There are 3 individual jira Mission , Expected to be 14 Days to complete , Results first person task 2 Weeks have passed and it is not over yet , Explain that there is a problem .

For this kind of , It means that the task is too challenging , Just give him more external work orders to exercise , Make time for other team members , Work overtime at night / Be on duty , They also arranged for him to come more . Work orders shall be clearly distinguished as far as possible , Specify by filling in the mandatory options , Instead of filling in the notes , Many people are too lazy to write in the notes .