enterprise replication log usage
Posted in 2011
Topics: High Availability & Replication, Storage & Space Management, Platform-Specific Issues, Versions, Editions & End-of-Life
IBM Informix Dynamic Server Version 11.50.FC6
HP-UX wuasp132 B.11.23 U 9000/800
enterprise replication in use
Question
How is data in blob space logged?
On our primary server we had a big job running inserting data into a
blobspace. I did not notice excessive log use.
However on the replicated server there was a huge amount of logs used.
Why is it that on the primary server not many logs were used, yet on the
destination server many logs were used - more that twice the number.
Next question
How can I tell what tables the data in the blob space belongs to?
I can get this when running dbschema
For example 2 lines of the schema
create table "paqis".history_xml
xml text in blobspc1 not null,
However there are serveral tables writing data to the blob space
I am running a BLOBSpace Report for res_live:paqis.history_xml
- been going 20 mins I suspect it will come back with a huge amount of
information rather than a summary
my oncheck came back with error after printing
4750 lines
ERROR: TBLspace number is invalid '0x2'.
Any ideas on this please ?
Log usage at target server is due to queuing blob data into receive queue
smartblob space. Looks like sbspace configured for CDR_QDATA_SBSPACE is a
logging sbspace.
At source server, ER do not queue blob space data into sendq (exception is
partition blob data). Data is directly read by network send thread from
the blob space. This is the reason for minimal log usage at source server.
One way to identify the table is by looking at 'onstat -g rcv full' output
at target server. This command prints last txn commit time at local server
and at source server for each replicate id. By comparing these timestamps
may be you can identify the table.
Replicate id to table name mapping can be found by looking at 'cdr list
repl' output or by looking at 'onstat -g cat repls' command output.
Thanks & Regards,
Nagaraju
From:
"KARL OLIVER" <karl.oliver@maf.govt.nz>
To:
ids@iiug.org
Date:
07/12/2011 09:17 PM
Subject:
enterprise replication log usage [24333]
IBM Informix Dynamic Server Version 11.50.FC6
HP-UX wuasp132 B.11.23 U 9000/800
enterprise replication in use
Question
How is data in blob space logged?
On our primary server we had a big job running inserting data into a
blobspace. I did not notice excessive log use.
However on the replicated server there was a huge amount of logs used.
Why is it that on the primary server not many logs were used, yet on the
destination server many logs were used - more that twice the number.
Next question
How can I tell what tables the data in the blob space belongs to?
I can get this when running dbschema
For example â 2 lines of the schema
create table "paqis".history_xml
xml text in blobspc1 not null,
However there are serveral tables writing data to the blob space
I am running a BLOBSpace Report for res_live:paqis.history_xml
- been going 20 mins I suspect it will come back with a huge amount of
information rather than a summary
*******************************************************************************
Forum Note: Use "Reply" to post a response in the discussion forum.
Karl,
We won't log the blobspace blob data when writing it, and also for a blob
space blob, we extract the blob when transmitting the replicated
transaction to the target server.
However, on the target, we may have to spool the blob because we can't
apply the replicated transaction until the complete transaction has been
received.
My guess is that your CDR_QUEUEMEM size is not large enough to hold the
transaction being received and thus we are having to spool the transaction
(and the LOB) in the sblob queue. And you configured the SBLOB space on
the target server as a logged SBLOB. Thus there would be an increase in
log consumption on the target. Not for the apply of the transaction, but
the logging performed when the LOB was spooled in the receive queue.
M.Pruet
|------------>
| From: |
|------------>
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|"KARL OLIVER" <karl.oliver@maf.govt.nz> |
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|------------>
| To: |
|------------>
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|ids@iiug.org |
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|------------>
| Date: |
|------------>
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|07/12/2011 08:17 PM |
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|------------>
| Subject: |
|------------>
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|enterprise replication log usage [24333] |
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|------------>
| Sent by: |
|------------>
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|ids-bounces@iiug.org |
>--------------------------------------------------------------------------------------------------------------------------------------------------|
IBM Informix Dynamic Server Version 11.50.FC6
HP-UX wuasp132 B.11.23 U 9000/800
enterprise replication in use
Question
How is data in blob space logged?
On our primary server we had a big job running inserting data into a
blobspace. I did not notice excessive log use.
However on the replicated server there was a huge amount of logs used.
Why is it that on the primary server not many logs were used, yet on the
destination server many logs were used - more that twice the number.
Next question
How can I tell what tables the data in the blob space belongs to?
I can get this when running dbschema
For example – 2 lines of the schema
create table "paqis".history_xml
xml text in blobspc1 not null,
However there are serveral tables writing data to the blob space
I am running a BLOBSpace Report for res_live:paqis.history_xml
- been going 20 mins I suspect it will come back with a huge amount of
information rather than a summary
*******************************************************************************
Forum Note: Use "Reply" to post a response in the discussion forum.
Madison
sorry cant read your reply
Nagaraju - not sure you are understanding my question.
I ran oncheck -pB res_live:history_xml - i know this is the table being
updated.
an oncheck -pe blobspc1
gives me what i want in regards to what is in the blob space
-------------------------------------------------------------------------------
OVERHEAD 19766
res_live:'paqis'.sample_history 894514
res_live:'paqis'.history_xml 2928606
FREE 1157114
Comes back with
ERROR: TBLspace number is invalid '0x2'.
I will log a call with IBM support about the error as it does not look right
Related threads
- Posting from the Informix-list
- Migrating from IDS 9.40.UC6 to 11.50.UC3
- Ip for a network session
- questions onstat -g