Understanding Alert Log

Hello

I am working on Aix 4.3 and Oracle7 Server Release 7.3.4.0.0 DB . The size is 30 Gb and would like to know how to Understand the Alert log . Any specific things that I should be look for daily . Any help would be appreciated .

Thanks

Afshin

Hi,
You dont have to understand alert log file on daily basis.Basically alert log file contains important error messages such as ORA-00600 ,deadlock error messages.Apart from this any changes made to the database by ALTER DATABASE or ALTER TABLESPACE command is recoreded in alert.log file.Also log switches and checkpoint intervals are recoreded in alert.log file.YOu can monitor alert.log file to see any ORA-00600 errors(bugs) are found in alert.log file and also to tune your checkpoint intervals and redo log switches you can see it.

In case of any help please be free to ask me at [email protected]

Regards.
Rohit Nirkhe,Oracle DBA,OCP 8i
[email protected]

I disagree. I personnaly think that every DBA should check the alert log at least once a day. The reasons for this many vary but any indications of errors should be addressed immediately rather than later.

Nizar

Quote:

---

Originally posted by nabaig
I disagree. I personnaly think that every DBA should check the alert log at least once a day. The reasons for this many vary but any indications of errors should be addressed immediately rather than later.

Nizar

---

I think it is up to the DBA. I administer 13 instances and I cannot really check all alert files daily. However, I check at least once a day the alerts of the PROD databases. Sometimes, 3-4 times a day. I usually open them in NOTEPAD and search for ORA- :-)

I have to disagree as well. I think the alert log needs to be reviewed daily. However, I also understand the time constriants placed on us by the normal daily life. I manage 40 databases spread over 12 servers, as well as spend time working with the developers to try and make my life easier. Also too much time is spent in meetings, but that's another entry entirely.

I use a set of scripts to review the alert logs daily and email me with any errors, warnings, etc. Basically the scripts can be set up to look for as much or as little as you want. It saves me time as I only have to review a smaller email for each database. I also want the scripts to email me a simple message if there is nothing to report. The scripts also rename the alert log so that I don't have to review a 100 Mbyte file if I want to go into it. I keep 30 days worth of alert logs online (yeah, I know. That's probably too much but it's my disk).

justdba,

do you mind to email me the scripts to send you error log through email ??

my email is [email protected]

thanks

I use OEM.

It inspects the alert log and sends an email whenever it finds something 'bad'.

Regards
Gert

Hi

I have read your reply. It is nice to read. It is interesting that,
your script reads the alert log and email to your address.
I would be pleased, if you send me the script. I would appreciate, if you could send me the steps, like how to extract from alert log and sending the information through email.

My email id [email protected]

Thanks

I have to disagree as well. It's recommended that you check your alert log at least once a day. But I made it my business to check it twice a day. If you don't understand the alert log how are you going to solve the error messages. ORA- 0600 is internal error that requires to call oracle support before you screw every thing.
If your alert log states that "checkpoint not complete"...Thread 1 cannot allocate new log, sequence 249.
If you don't understand the error stack how will you resolve the problem?
Be familiar with the alert log.
Thanks

I haven't looked at a production alert.log file in about 3 years. I would advise you to learn which entries in the alert.log file indicate "trouble" and then write some scripts to monitor for those trouble events. Monitoring of the alert.log script can be automated with simple OS tools and mail.

Your monitoring scripts don't have to give you every detail of what the problem is. If I get a page saying that I have an ORA-600, I will go into the alert.log and investigate.

Whenever I install a new version of Oracle (in test, of course), I monitor the alert.logs RELIGIOUSLY looking for new "trouble" events. If a new event is found, I update my scripts and never worry about it again.