/ hbayskate81's Library/ Notes/ Additionally to the database level, you can find also several standard instances at the host, operating system, and storage levels. Should you be not cautious, misoperations in the host network level may possibly also have a fatal impact around the system
Additionally to the database level, you can find also several standard instances at the host, operating system, and storage levels. Should you be not cautious, misoperations in the host network level may possibly also have a fatal impact around the system
from web site
Case Research
1. Method commands issued by error
HPUNIXOracle10.two. After logging in with root, I developed a brand new host user. I unwittingly knocked on a hostname–a. Do you know what occurred behind me?
It truly is confused with uname-a, hostname-a directly changed the host name to -a ...
The listener listens for the host name, and now the host cannot be discovered, continuous errors are reported, along with the background trc file also continuously reports errors. There are four situations on this host, and they can't be connected at the similar time ... Spectacular ... .. Quickly the log fills up the file system.
I could not find out the explanation, but identified that the file program was utilised really promptly, so I wanted to stop the library very first, and then verify the cause. Consequently, ora600 was began. Fortunately, the test environment database just isn't official. It is really memorable.
2. Switch production storage by mistake
A very cold operation. 1 afternoon last year, I utilized to switch HA on the disaster recovery side in the cabinet. I was dizzy and I pressed it. I manually switched the production side on the cabinet. More than 20 database systems have been operating on it. ... the consequences are unimaginable. Luckily, I swiftly switched back, pretending to become absolutely nothing, and my hands have been shaking ... Afterwards, I secretly asked if there had been any abnormalities in some on the net service systems. MM just said some The minutes are very slow, the database doesn't respond, and then it is actually normal once again, Khan ...
From now on, there is a feeling of 'you are the one' for anyone who is in the production environment.
three. Storage maintenance harmful misuse
Around the storage navsphere management interface of cx700, configure a storage. A colleague took more than and opened one more IE window of your storage in the production environment. I took more than once again. When I saw the configuration of this storage as if I opened it, I started to delete the storagegroup. Luckily, another colleague next to me saw that the host name was wrong, which prevented me from continuing to delete it (I explained the methods to configure storage to him and started the operation).
Delete the lun and drop the CRM data of the production atmosphere.
This matter is terrible, and men and women have been in a poor state that day. The more you do factors inside the future, the much more you realize that the state just isn't good, the extra you have to be a lot more cautious. There's also a relative path to delete files just before, ../path way, the oracle system with the test environment was accidentally deleted, plus the absolute path might be used inside the future.
4. Delete operating system files by mistake
Once installed RAC on IBMp570, resulting from buyer network problems, the outcome failed. When deleting RAC, rm-inittab * .crsd and a number of other RAC startup files, I didn't pay attention to delete a file of AIX, as well as the method couldn't begin Too. Later, due to IBM engineers who restored the technique. As a result, function was only closed at 3 pm.
five. Misoperation to execute system commands
After adding nodes inside the production environment and staying for two days and two nights, my colleagues executed pvid = yes on the production machine, which triggered information loss, and ultimately fought to reinstall RAC for two days.
Precautions
1. recover partition table and database user are strictly separated
Within a production environment, DBA need to not be given root privileges to stop the effect of the operation on the entire program. Even when the DBA might know the method effectively, the skilled division of labor calls for a system administrator to perform system-level upkeep.
how to recover gpt partition table of method failures caused by improper DBA operations.
two. It is all about storage
The storage finally includes each of the user's information, so any operation for storage can not be sloppy. When adding or removing difficult disks and formatting partitions, it truly is necessary to strictly perform disk confirmation and partition comparison to prevent 'dumping the salary' as a result of misuse.
3. Power is Energy
The power supply is Energy, which can be the supply of all energy, so when the power provide is interrupted, all environments from the technique may possibly be affected. When coping with energy complications, care really should be taken, as there are plenty of circumstances exactly where the database can't be began as a result of energy failure. Do not let the database crash on account of energy challenges.
The above content material is excerpted from Gai Guoqiang's 'OracleDBA Notes 4 Information Security Warning Record'.
1. Method commands issued by error
HPUNIXOracle10.two. After logging in with root, I developed a brand new host user. I unwittingly knocked on a hostname–a. Do you know what occurred behind me?
It truly is confused with uname-a, hostname-a directly changed the host name to -a ...
The listener listens for the host name, and now the host cannot be discovered, continuous errors are reported, along with the background trc file also continuously reports errors. There are four situations on this host, and they can't be connected at the similar time ... Spectacular ... .. Quickly the log fills up the file system.
I could not find out the explanation, but identified that the file program was utilised really promptly, so I wanted to stop the library very first, and then verify the cause. Consequently, ora600 was began. Fortunately, the test environment database just isn't official. It is really memorable.
2. Switch production storage by mistake
A very cold operation. 1 afternoon last year, I utilized to switch HA on the disaster recovery side in the cabinet. I was dizzy and I pressed it. I manually switched the production side on the cabinet. More than 20 database systems have been operating on it. ... the consequences are unimaginable. Luckily, I swiftly switched back, pretending to become absolutely nothing, and my hands have been shaking ... Afterwards, I secretly asked if there had been any abnormalities in some on the net service systems. MM just said some The minutes are very slow, the database doesn't respond, and then it is actually normal once again, Khan ...
From now on, there is a feeling of 'you are the one' for anyone who is in the production environment.
three. Storage maintenance harmful misuse
Around the storage navsphere management interface of cx700, configure a storage. A colleague took more than and opened one more IE window of your storage in the production environment. I took more than once again. When I saw the configuration of this storage as if I opened it, I started to delete the storagegroup. Luckily, another colleague next to me saw that the host name was wrong, which prevented me from continuing to delete it (I explained the methods to configure storage to him and started the operation).
Delete the lun and drop the CRM data of the production atmosphere.
This matter is terrible, and men and women have been in a poor state that day. The more you do factors inside the future, the much more you realize that the state just isn't good, the extra you have to be a lot more cautious. There's also a relative path to delete files just before, ../path way, the oracle system with the test environment was accidentally deleted, plus the absolute path might be used inside the future.
4. Delete operating system files by mistake
Once installed RAC on IBMp570, resulting from buyer network problems, the outcome failed. When deleting RAC, rm-inittab * .crsd and a number of other RAC startup files, I didn't pay attention to delete a file of AIX, as well as the method couldn't begin Too. Later, due to IBM engineers who restored the technique. As a result, function was only closed at 3 pm.
five. Misoperation to execute system commands
After adding nodes inside the production environment and staying for two days and two nights, my colleagues executed pvid = yes on the production machine, which triggered information loss, and ultimately fought to reinstall RAC for two days.
Precautions
1. recover partition table and database user are strictly separated
Within a production environment, DBA need to not be given root privileges to stop the effect of the operation on the entire program. Even when the DBA might know the method effectively, the skilled division of labor calls for a system administrator to perform system-level upkeep.
how to recover gpt partition table of method failures caused by improper DBA operations.
two. It is all about storage
The storage finally includes each of the user's information, so any operation for storage can not be sloppy. When adding or removing difficult disks and formatting partitions, it truly is necessary to strictly perform disk confirmation and partition comparison to prevent 'dumping the salary' as a result of misuse.
3. Power is Energy
The power supply is Energy, which can be the supply of all energy, so when the power provide is interrupted, all environments from the technique may possibly be affected. When coping with energy complications, care really should be taken, as there are plenty of circumstances exactly where the database can't be began as a result of energy failure. Do not let the database crash on account of energy challenges.
The above content material is excerpted from Gai Guoqiang's 'OracleDBA Notes 4 Information Security Warning Record'.
