Brunzema, Martin schrieb:
-----Original Message-----
From: Michael Jürgens [mailto:[EMAIL PROTECTED]
Sent: Tuesday, September 05, 2006 2:37 PM
To: Brunzema, Martin
Subject: Re: Savepoint Question
Brunzema, Martin schrieb:
-----Original Message-----
I?ve a question about savepoints.
According to maxdb doc the database will force a savepoint
when the log
volume filling level reach 66% and also before a data backup
started -
right?
Since few days this will only happen before database backup.
The log volume use reaches 85% and more and falls down
to 5% when
databackup occurs.
What happens there?
Hi Michael,
did you made any changes to parameters or did you upgrade
your software?
Which software version do you use?
Do you get into the situation of "log full" if you do not
start the
databackup?
kind regards, Martin
Hi Martin,
we are now using maxDB 7.5.0.30
At nearly the same time we switched to new Hardware.
Parameters:
old new
Version 7.5.0.24 -> 7.5.0.30
LogSpace 100MB -> 8 GB
I can´t test the "log full" situation, because it´s in production.
Also autolog will not perform backups and manual
logbackup does not
reduce log volume usage.
Hi Michael,
could you please check whether there are error messages in
the file knldiag
resp. knldiag.err concerning the logbackup. E.g. the
autosave terminates
with an error immediately after switching it on.
There are no failures.
After swiching off and on autlog saves two autolog files with
1398341632
and 1221435392 Bytes, but "info log" returns nearly
the same filling level (see on bottom of this email):
Name |
Value
....
Hi Michael,
the values are looking quite good. Before the logbackup the "used size" is
determined
by the number of unsaved pages:
Used Size (KB) = 3000144
Used Size (%) = 27
Not Saved (KB) = 3000144
Not Saved (%) = 27
after the backup the number of unsaved pages has been lowered so that the last
savepoint
determines the size of the used log (all pages younger than the last savepoint
are needed
for the next restart):
Used Size (KB) = 3018552
Used Size (%) = 27
Not Saved (KB) = 1655960
Not Saved (%) = 15
If the value of "not saved" exceeds the backup segment size then the automatic
logbackup
should start. The value of "used size" will increase until the next savepoint
is written.
This will be at the point when this value exceeds 66 %.
I'm a little bit confused about your statement that the autosave save two files
with different
size. Is it possible that you meant the manually started logbackup. For enabled
autosave you should
find something like
2006-09-05 15:03:45 0x1080 52104 AUTOSAVE standby mode on
in the file knldiag. Any subsequent messages regarding the autosave will also
have the label
AUTOSAVE.
kind regards, Martin
Hi Martin,
the problem still exists.
I´ve looked today at "info log" and "not saved" is greater then "backup
Segment size":
Name |
Value
Log Mirrored =
NO
Log Writing = ON
Log Automatic Overwrite = OFF
Max. Size (KB) = 11261472
Backup Segment Size (KB) = 1365328
Used Size (KB) = 1608368
Used Size (%) = 14
Not Saved (KB) = 1608368
Not Saved (%) = 14
Log Since Last Data Backup (KB) = 0
Savepoints = 98932
Checkpoints = 0
Physical Reads = 1821375
Physical Writes = 195890076
Queue Size (KB) = 400
Queue Overflows = 0
Group Commits = 3076066
Waits for Logwriter = 198390589
Max. Waits = 35
Average Waits = 0
After stopping and starting autolog I found the following in the knldiag
file:
2006-09-06 09:00:11 15 52029 AUTOSAVE standby mode off
.....
2006-09-06 09:00:54 15 52104 AUTOSAVE standby mode on
2006-09-06 09:00:54 15 11560 COMMUNIC Releasing T57
2006-09-06 09:00:54 15 12929 TASKING Task T57 started
2006-09-06 09:00:54 15 11007 COMMUNIC wait for connection T57
2006-09-06 09:00:54 16 52030 AUTOSAVE started
2006-09-06 09:00:54 303 12821 TASKING Thread 303 starting
2006-09-06 09:00:54 303 11597 IO Open
'/home/dbbackup/PROD_R2/logsave/logsave.096' successfull, fd: 38
2006-09-06 09:00:54 303 11565 startup DEVi started
2006-09-06 09:00:54 16 11000 vasynope
'/home/dbbackup/PROD_R2/logsave/logsave.096' devno 32 T48 succeeded
2006-09-06 09:00:54 16 52101 AUTOSAVE Filetype: file
2006-09-06 09:00:54 16 52024 AUTOSAVE 200 pages ->
"OD_R2/logsave/logsave.096"
2006-09-06 09:00:56 19 WNG 11824 COMMUNIC Releasing T66 command
timeout
2006-09-06 09:00:56 19 12929 TASKING Task T66 started
2006-09-06 09:00:56 19 11007 COMMUNIC wait for connection T66
2006-09-06 09:00:59 16 52024 AUTOSAVE 41600 pages ->
"_R2/logsave/logsave.096"
2006-09-06 09:01:04 16 52024 AUTOSAVE 73600 pages ->
"_R2/logsave/logsave.096"
2006-09-06 09:01:09 16 52024 AUTOSAVE 108800 pages ->
"R2/logsave/logsave.096"
2006-09-06 09:01:14 16 52024 AUTOSAVE 141800 pages ->
"R2/logsave/logsave.096"
2006-09-06 09:01:18 16 52024 AUTOSAVE 170680 pages ->
"R2/logsave/logsave.096"
2006-09-06 09:01:18 16 11000 vasynclo
'/home/dbbackup/PROD_R2/logsave/logsave.096' devno 32 T48
2006-09-06 09:01:18 10 12822 TASKING Thread 303 joining
2006-09-06 09:01:18 303 11566 stop DEVi stopped
2006-09-06 09:01:19 16 52031 AUTOSAVE ready
It seems, that autosave don´t wakeup from standby mode.
It is possible, that this is problem with system tables.
I don´t know exactly if, I reinstall the system tables after upgrade
from 7.5.0.24 to 7.5.0.30.
regards,
Michael
--
MaxDB Discussion Mailing List
For list archives: http://lists.mysql.com/maxdb
To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
