Chunck writes
Hello!
What can be a reason for big chunck writes.
Here is output with onstat -F|more.
Fg Writes LRU Writes Chunk Writes
0 0 747145
address flusher state data
7000000d04f3850 0 I 0 = 0X0
7000000d04f4078 1 I 0 = 0X0
7000000d04f48a0 2 I 0 = 0X0
7000000d04f50c8 3 I 0 = 0X0
7000000d04f58f0 4 I 0 = 0X0
7000000d04f6118 5 I 0 = 0X0
7000000d04f6940 6 I 0 = 0X0
7000000d04f7168 7 I 0 = 0X0
7000000d04f7990 8 I 0 = 0X0
7000000d04f81b8 9 I 0 = 0X0
7000000d04f89e0 10 I 0 = 0X0
7000000d04f9208 11 I 0 = 0X0
7000000d04f9a30 12 I 0 = 0X0
7000000d04fa258 13 I 0 = 0X0
7000000d04faa80 14 I 0 = 0X0
7000000d04fb2a8 15 I 0 = 0X0
7000000d04fbad0 16 I 0 = 0X0
Hi,
chunk writes are checkpoint writes and not bad (unless
the checkpoint durations are very high - see message log).
Possible reasons for high chunk writes:
- short checkpoint intervalls
- small PHYSFILE (a small physical logspace triggers checkpoints)
- high LRU_MAX_DIRTY value or little LRUs
- huge BUFFER pool
Regards,
Andreas Kutsche
> -----Ursprüngliche Nachricht-----
> Von: forum.subscriber@iiug.org [mailto:forum.subscriber@iiug.org]Im
> Auftrag von Leon
> Gesendet: Dienstag, 19. April 2005 10:35
> An: ids@iiug.org
> Betreff: Chunck writes [4747]
>
>
> Hello!
>
> What can be a reason for big chunck writes.
> Here is output with onstat -F|more.
>
> Fg Writes LRU Writes Chunk Writes
> 0 0 747145
>
> address flusher state data
> 7000000d04f3850 0 I 0 = 0X0
> 7000000d04f4078 1 I 0 = 0X0
> 7000000d04f48a0 2 I 0 = 0X0
> 7000000d04f50c8 3 I 0 = 0X0
> 7000000d04f58f0 4 I 0 = 0X0
> 7000000d04f6118 5 I 0 = 0X0
> 7000000d04f6940 6 I 0 = 0X0
> 7000000d04f7168 7 I 0 = 0X0
> 7000000d04f7990 8 I 0 = 0X0
> 7000000d04f81b8 9 I 0 = 0X0
> 7000000d04f89e0 10 I 0 = 0X0
> 7000000d04f9208 11 I 0 = 0X0
> 7000000d04f9a30 12 I 0 = 0X0
> 7000000d04fa258 13 I 0 = 0X0
> 7000000d04faa80 14 I 0 = 0X0
> 7000000d04fb2a8 15 I 0 = 0X0
> 7000000d04fbad0 16 I 0 = 0X0
>
>
>
Leon
Checkpoints !!
Chunk Writes are good. It means all you disk writes are being done in the
most efficient way during your checkpoints. Provided your checkpoints are
in the order of 2-3 seconds you have nothing to worry about, you have a
reasonably well configured system.
Keith
-> -----Original Message-----
-> From: forum.subscriber@iiug.org [mailto:forum.subscriber@iiug.org]On
-> Behalf Of Leon
-> Sent: Tuesday, April 19, 2005 9:35 AM
-> To: ids@iiug.org
-> Subject: Chunck writes [4747]
->
->
-> Hello!
->
-> What can be a reason for big chunck writes.
-> Here is output with onstat -F|more.
->
-> Fg Writes LRU Writes Chunk Writes
-> 0 0 747145
->
-> address flusher state data
-> 7000000d04f3850 0 I 0 = 0X0
-> 7000000d04f4078 1 I 0 = 0X0
-> 7000000d04f48a0 2 I 0 = 0X0
-> 7000000d04f50c8 3 I 0 = 0X0
-> 7000000d04f58f0 4 I 0 = 0X0
-> 7000000d04f6118 5 I 0 = 0X0
-> 7000000d04f6940 6 I 0 = 0X0
-> 7000000d04f7168 7 I 0 = 0X0
-> 7000000d04f7990 8 I 0 = 0X0
-> 7000000d04f81b8 9 I 0 = 0X0
-> 7000000d04f89e0 10 I 0 = 0X0
-> 7000000d04f9208 11 I 0 = 0X0
-> 7000000d04f9a30 12 I 0 = 0X0
-> 7000000d04fa258 13 I 0 = 0X0
-> 7000000d04faa80 14 I 0 = 0X0
-> 7000000d04fb2a8 15 I 0 = 0X0
-> 7000000d04fbad0 16 I 0 = 0X0
->
->
->
********************************************************************************
**
This message is sent in strict confidence for the addressee only. It may
contain legally privileged information. The contents are not to be disclosed
to anyone other than the addressee. Unauthorised recipients are requested
to preserve this confidentiality and to advise the sender immediately of any
error in transmission.
This footnote also confirms that this email message has been swept for the
presence of computer viruses, however we cannot guarantee that this message
is free from such problems.
********************************************************************************
**
Although
I don't do my own performance tuning (someone else much more
skilled in the mystical arts does this...), I use this little puppy for
a quick look at the engine logs to extract the checkpoint counts and
"average" checkpoint. We have guidelines for when to call the "Defense
Against The Dark Arts" teacher for performance tuning.
Caution: Check for word wrap if you copy/paste this and try to use it!
---Begin Snippet---
# *** extract a list of seconds from the enginlogs, get checkpoint count
grep Checkpoint ${ENGINELOGS}/enginelog.* | awk '{print $6 }' >
${TMP}/sec.txt
sec_cnt=`cat ${TMP}/sec.txt | wc -l`
$INFORMIXBIN/isql $DBNAME << !!
# *** make a temp table to check the time average
create temp table tmp_sec$$
(sec integer) with no log;
load from ${TMP}/sec.txt insert into tmp_sec$$;
# *** average second and load count check
select count(*) sec_cnt, "${sec_cnt}", avg(sec)
from tmp_sec$$
group by 2;
drop table tmp_sec$$;!!
---End Snippet---
Rob K
-----Original Message-----
From: forum.subscriber@iiug.org [mailto:forum.subscriber@iiug.org] On
Behalf Of Simmons, Keith
Sent: Tuesday, April 19, 2005 7:06 AM
To: ids@iiug.org
Subject: RE: Chunck writes [4752]
Leon
Checkpoints !!
Chunk Writes are good. It means all you disk writes are being done in
the
most efficient way during your checkpoints. Provided your checkpoints
are
in the order of 2-3 seconds you have nothing to worry about, you have a
reasonably well configured system.
Keith
-> -----Original Message-----
-> From: forum.subscriber@iiug.org [mailto:forum.subscriber@iiug.org]On
-> Behalf Of Leon
-> Sent: Tuesday, April 19, 2005 9:35 AM
-> To: ids@iiug.org
-> Subject: Chunck writes [4747]
->
->
-> Hello!
->
-> What can be a reason for big chunck writes.
-> Here is output with onstat -F|more.
->
-> Fg Writes LRU Writes Chunk Writes
-> 0 0 747145
->
-> address flusher state data
-> 7000000d04f3850 0 I 0 = 0X0
-> 7000000d04f4078 1 I 0 = 0X0
-> 7000000d04f48a0 2 I 0 = 0X0
-> 7000000d04f50c8 3 I 0 = 0X0
-> 7000000d04f58f0 4 I 0 = 0X0
-> 7000000d04f6118 5 I 0 = 0X0
-> 7000000d04f6940 6 I 0 = 0X0
-> 7000000d04f7168 7 I 0 = 0X0
-> 7000000d04f7990 8 I 0 = 0X0
-> 7000000d04f81b8 9 I 0 = 0X0
-> 7000000d04f89e0 10 I 0 = 0X0
-> 7000000d04f9208 11 I 0 = 0X0
-> 7000000d04f9a30 12 I 0 = 0X0
-> 7000000d04fa258 13 I 0 = 0X0
-> 7000000d04faa80 14 I 0 = 0X0
-> 7000000d04fb2a8 15 I 0 = 0X0
-> 7000000d04fbad0 16 I 0 = 0X0
->
->
->
************************************************************************
**********
This message is sent in strict confidence for the addressee only. It
may
contain legally privileged information. The contents are not to be
disclosed
to anyone other than the addressee. Unauthorised recipients are
requested
to preserve this confidentiality and to advise the sender immediately of
any
error in transmission.
This footnote also confirms that this email message has been swept for
the
presence of computer viruses, however we cannot guarantee that this
message
is free from such problems.
************************************************************************
**********
It just means that all of your writes are being done at checkpoint time and
none
are being forced by LRU Queue's filling with dirty buffers beyond the level set
in MRU_MAX_DIRTY. If your checkpoints are not long this is OK. Reducing
LRU_MAX/MIN_DIRTY will shift some of that IO work to LRU Writes and reduce
checkpoint durations, but if they are already acceptably low, then Chunk Writes
are more efficient in resource usage. For most OLTP type servers things hum
best with about 60-70% LRU Writes and 40-30% Chunk Writes but YMMV and if you
are not having problems with apps pausing during checkpoints, this is good.
Art S. Kagel
----- Original Message -----
From: Leon <leon.pobor@hpb.hr>
At: 4/19 6:01
> Hello!
>
> What can be a reason for big chunck writes.
> Here is output with onstat -F|more.
>
> Fg Writes LRU Writes Chunk Writes
> 0 0 747145
>
> address flusher state data
> 7000000d04f3850 0 I 0 = 0X0
> 7000000d04f4078 1 I 0 = 0X0
> 7000000d04f48a0 2 I 0 = 0X0
> 7000000d04f50c8 3 I 0 = 0X0
> 7000000d04f58f0 4 I 0 = 0X0
> 7000000d04f6118 5 I 0 = 0X0
> 7000000d04f6940 6 I 0 = 0X0
> 7000000d04f7168 7 I 0 = 0X0
> 7000000d04f7990 8 I 0 = 0X0
> 7000000d04f81b8 9 I 0 = 0X0
> 7000000d04f89e0 10 I 0 = 0X0
> 7000000d04f9208 11 I 0 = 0X0
> 7000000d04f9a30 12 I 0 = 0X0
> 7000000d04fa258 13 I 0 = 0X0
> 7000000d04faa80 14 I 0 = 0X0
> 7000000d04fb2a8 15 I 0 = 0X0
> 7000000d04fbad0 16 I 0 = 0X0