Re: Solution: Help with full logical logs
Posted in 1998
Topics: Backup & Restore, Server Administration, Logging & Checkpoints
This is a multi-part message in MIME format.
--------------3B64F00B78EF4FE80E2747BA
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
"Art S. Kagel" wrote:
>
> Joachim Scherer wrote:
> >
> > Hi:
> > my logical logs are full. ontape -a doesn't work anymore. How can free
> > the logs without loosing my data. And hopefully without doing a restore.
>
> Right, ontape -a will not work if the last log is full because it
> cannot log the fact of the log backup. I guess you have LBU_PRESERVE
> set to 0 or not present in your ONCONFIG. The only option is to change
> LTAPEDEV to /dev/null in onmonitor (changing it in the file is not
> enough it has to be changed in memory unlike TAPEDEV so you need to use
> onmonitor).
Changing LTAPEDEV in onmonitor to /dev/null did not work. I then changed
it in $informix/etc/onconfig.ifmx, stoped and startd the engine. That worked.
Now onstat -l shows:
# onstat -l
INFORMIX-OnLine Version 7.12.UC1 -- On-Line -- Up 00:17:20 -- 8976 Kbytes
Physical Logging
Buffer bufused bufsize numpages numwrits pages/io
P-1 0 16 2 1 2,00
phybegin physize phypos phyused %used
10003f 500 381 0 0,00
Logical Logging
Buffer bufused bufsize numrecs numpages numwrits recs/pages pages/io
L-1 0 16 8 7 3 1,1 2,3
address number flags uniqid begin size used %used
a050ce8 1 F------ 0 100233 250 0 0,00
a050d04 2 F------ 0 10032d 250 0 0,00
a050d20 3 U-B---- 33 100427 250 250 100,00
a050d3c 4 U---C-L 34 100521 250 7 2,80
a050d58 5 F------ 0 10061b 250 0 0,00
a050d74 6 F------ 0 100715 250 0 0,00
But still ontape doesn't work!! I still get the same error message
(Server in incompatible state)
which points to a post from June Tong in this thread.
>
> This will cause all logs to be wiped clean without saving them. For
> safety IMMEDIATELY after the logs clear up run ontape -s -L 0 to get
> a FULL backup since you have just thrown your log files away (you can
> probably get away with a -L 1 or -L 2 if the level 0 is very recent but
> I like to be safe and use -L0 after a near disaster). After that
> change LBU_PRESERVE to 1 in your ONCONFIG file so that the last log
> file cannot fill up then you will be able to us ontape -a to recover
> next time. Don't forget to change LTAPEDEV back and to bounce the
> server when the archive is complete to get the new LBU_PRESERVE value
> activated.
>
Ok now I have set LBU_PRESERVE to 1. Why is the standard setting 0?
It would be much safer (especially for people like me) to have that
variable set to 1.
Well, this goes out to the informix people.
> Art S. Kagel
>
> PS- Be thankful you are not using Sybase where this problem means doing
> a full restore to recover.
I'm also thankfull to have comp.databases.informix ;-)
Joachim Scherer
--------------3B64F00B78EF4FE80E2747BA
Content-Type: text/x-vcard; charset=us-ascii;
name="joachim.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Joachim Scherer
Content-Disposition: attachment;
filename="joachim.vcf"
begin:vcard
n:Scherer;Joachim
x-mozilla-html:TRUE
version:2.1
email;internet:joachim@diana.scientific.de
x-mozilla-cpt:;0
fn:Joachim Scherer
end:vcard
--------------3B64F00B78EF4FE80E2747BA--