strange "Log Full"

[Fan Ruo Xin]

Last monday, I tried to create a working table, and failed because of 'Log
Full'. There were two applications running at that time - one is autoload
(during the split phase), another one is "insert into a (NLI) ... SELECT ...
FROM ...".
I checked the db2diag.log (level=4) and notification file. It was
interesting that there was no any information about log full.
I had to kill the autoload process.
I haven'g got time to see if I can reproduce it on the testing machine. Have
some one met the same scenario? Are you lucky that you have some idea about
this?
TIA

[Larry]

Wonder if your insert stmt could have been a long-running insert where
it hadn't committed yet ... but was in the process of filling up the
log. Perhaps it never got to the point where it put any msgs out because
it never committed?

Larry Edelstein

Fan Ruo Xin wrote:

Last monday, I tried to create a working table, and failed because of 'Log
Full'. There were two applications running at that time - one is autoload
(during the split phase), another one is "insert into a (NLI) ... SELECT ...
FROM ...".
I checked the db2diag.log (level=4) and notification file. It was
interesting that there was no any information about log full.
I had to kill the autoload process.
I haven'g got time to see if I can reproduce it on the testing machine. Have
some one met the same scenario? Are you lucky that you have some idea about
this?
TIA

[Fan Ruo Xin]

Thanks, Larry.
My insert stmt should not cause the problem. Because I activated the Not
Logged Initially.

"Larry" <La***@nospam.net> wrote in message
news:cD********************@news4.srv.hcvlny.cv.ne t...

Wonder if your insert stmt could have been a long-running insert where
it hadn't committed yet ... but was in the process of filling up the
log. Perhaps it never got to the point where it put any msgs out because
it never committed?

Larry Edelstein

Fan Ruo Xin wrote:

Last monday, I tried to create a working table, and failed because of 'Log Full'. There were two applications running at that time - one is autoload (during the split phase), another one is "insert into a (NLI) ... SELECT .... FROM ...".
I checked the db2diag.log (level=4) and notification file. It was
interesting that there was no any information about log full.
I had to kill the autoload process.
I haven'g got time to see if I can reproduce it on the testing machine. Have some one met the same scenario? Are you lucky that you have some idea about this?
TIA

[Mark Yudkin]

I agree with Larry. NOT LOGGED INITIALLY only refers to INSERTs in the same
unit of work as the CREATE that specifies the no logging.

"Fan Ruo Xin" <fa*****@sbcglobal.net> wrote in message
news:LI***************@newssvr32.news.prodigy.com. ..

Thanks, Larry.
My insert stmt should not cause the problem. Because I activated the Not
Logged Initially.

"Larry" <La***@nospam.net> wrote in message
news:cD********************@news4.srv.hcvlny.cv.ne t...

Wonder if your insert stmt could have been a long-running insert where
it hadn't committed yet ... but was in the process of filling up the
log. Perhaps it never got to the point where it put any msgs out because
it never committed?

Larry Edelstein

Fan Ruo Xin wrote:

Last monday, I tried to create a working table, and failed because of 'Log Full'. There were two applications running at that time - one is autoload (during the split phase), another one is "insert into a (NLI) ...

SELECT

... FROM ...".
I checked the db2diag.log (level=4) and notification file. It was
interesting that there was no any information about log full.
I had to kill the autoload process.
I haven'g got time to see if I can reproduce it on the testing

machine.

Have some one met the same scenario? Are you lucky that you have some idea about this?
TIA

[Knut Stolze]

Fan Ruo Xin wrote:

My insert stmt should not cause the problem. Because I activated the Not
Logged Initially.

Do you have auto-commit turned off?
Did you run an ALTER TABLE ... ACTIVATE NOT LOGGED INITIALLY in the same
transaction/unit-of-work in which the INSERT is run? If not, then you
don't have NLI for the insert.

--
Knut Stolze
Information Integration
IBM Germany / University of Jena

[Fan Ruo Xin]

Thanks Mark & Knut.
Yes I am 100% sure that they are in the same transaction. I thought I got
wrong db2 sql error message. If my log full, the information should be
recorded in db2 diaglog.

"Knut Stolze" <st****@de.ibm.com> wrote in message
news:cc**********@fsuj29.rz.uni-jena.de...

Fan Ruo Xin wrote:

My insert stmt should not cause the problem. Because I activated the Not
Logged Initially.

Do you have auto-commit turned off?
Did you run an ALTER TABLE ... ACTIVATE NOT LOGGED INITIALLY in the same
transaction/unit-of-work in which the INSERT is run? If not, then you
don't have NLI for the insert.

--
Knut Stolze
Information Integration
IBM Germany / University of Jena

[Knut Stolze]

Fan Ruo Xin wrote:

Thanks Mark & Knut.
Yes I am 100% sure that they are in the same transaction. I thought I got
wrong db2 sql error message. If my log full, the information should be
recorded in db2 diaglog.

I would think that you should have gotten an error message in the
db2diag.log.

Oh, are you logging onto the same file system on which the db2diag.log is
also written? Maybe you ran into a disk-full situation and DB2 couldn't
even write the db2diag.log any more?

--
Knut Stolze
Information Integration
IBM Germany / University of Jena