IDS Forum

Re: ONCONFIG comments

Posted By: Jack Parker
Date: Wednesday, 11 May 2005, at 5:35 a.m.

In Response To: ONCONFIG comments (Bono Vox)

Comments embedded.

If somebody told me to change my $ONCONFIG, I'd take it with a large grain
of salt and check up on every parameter they suggested I modify. I'd then
make the changes one at a time and see how each one worked differently.

Looks like an OLTP instance. Also looks like you're sharing the box with
something else? Doesn't quite make sense, you've got 12GB of memory and
connections for 1500 users, and yet you appear to be only allocating 500MB
or so for the instance.

j.
----- Original Message -----
From: "Bono Vox" <in4mixperu@hotmail.com>
To: <ids@iiug.org>
Sent: Tuesday, May 10, 2005 4:00 AM
Subject: ONCONFIG comments [4891]


[...]

> ============================================================
> HP-UX 11.11 (64 bits)
> 4 CPU's
> 12 GB Memory
> Array VA7400
> Informix 7.31 FD8
> ============================================================
>
> ONCONFIG
> =======
>
> # Root Dbspace Configuration
>
> ROOTNAME rootdbs # Root dbspace name
> ROOTPATH /dev/informix/nazcasysch01 # Path for device containing
> root dbspace

I trust this is a symbolic link.

> ROOTOFFSET 0 # Offset of root
> dbspace into device (Kbytes)

I've never been keen on offset 0, some drives consider the first couple of
KB as their own - I have seen this wipe out instances in the past. I
usually go for an offset of 40.

> ROOTSIZE 204800 # Size of root
dbspace
> (Kbytes)

Seems a little small, but if it is sufficient...

>
> # Disk Mirroring Configuration Parameters
>
> MIRROR 1 # Mirroring flag (Yes = 1, No = 0)
> MIRRORPATH # Path for device containing mirrored root
> MIRROROFFSET 0 # Offset into mirrored device (Kbytes)

??? You have mirroring turned on, but no device specified to mirror to. I
would think this would generate an error during instance startup.

>
> # Physical Log Configuration
>
> PHYSDBS phydbs # Location (dbspace) of physical log
> PHYSFILE 64000 # Physical log file size (Kbytes)

I would check the sizing of this by checking Checkpoint frequency in your
log.

>
> # Logical Log Configuration
>
> LOGFILES 80 # Number of logical log files
> LOGSIZE 32768 # Logical log size (Kbytes)

Quite a few of these. Does this indicate that you aren't backing them off
as fast as they fill? How many are full when the backup comes to town?

[...]

> TBLSPACE_STATS 1

If you are using table space statistics then this is fine, but othetrwise it
is overhead and you might want to go back to 0.

>
> # System Archive Tape Device
>
> TAPEDEV /dev/rmt/2m # Tape device path
> #TAPEDEV /dev/rmt/2m # Tape device path
> TAPEBLK 4000 # Tape block size (Kbytes)
> TAPESIZE 40000000 # Maximum amount of data to put
> on tape (Kbytes)

Doesn't look like a no-rewind device, are you using this for backups?

>
> # Log Archive Tape Device
>
> LTAPEDEV /dev/rmt/2m # Log tape device path
> #LTAPEDEV /dev/rmt/2m # Log tape device path
> LTAPEBLK 4000 # Log tape block size (Kbytes)
> LTAPESIZE 40000000 # Max amount of data to put on log tape
> (Kbytes)

Looks like the same device. Are you using logging?

[...]

> NETTYPE ipcshm,3,250,CPU # Configure poll thread(s) for
nettype
> NETTYPE soctcp,3,250,CPU # Configure poll thread(s) for
nettype

While of minimal impact, still this is 6 processes to handle user
connections, 3 shared memory and 3 network. Check how many users you
actually have and adjust accordingly.

> DEADLOCK_TIMEOUT 60 # Max time to wait of lock in
> distributed env.
> RESIDENT 0 # Forced residency flag (Yes =
> 1, No = 0)

Not resident? Check release notes and see if there is an issue with
residency on this platform.

>
> MULTIPROCESSOR 1 # 0 for single-processor, 1 for
> multi-processor
> NUMCPUVPS 3 # Number of user (cpu) vps

If this were Solaris, I'd bump this to 4.

> SINGLE_CPU_VP 0 # If non-zero, limit number of cpu vps
> to one
>
> NOAGE 0 # Process aging
> AFF_SPROC 0 # Affinity start processor
> AFF_NPROCS 0 # Affinity number of processors

I've forgotten if NOAGE should be 1 or 0 - again this is something you'd
check release notes for. You might also want to affinitize the processors.

>
> # Shared Memory Parameters
>
> LOCKS 1024000 # Maximum number of locks
> BUFFERS 850000 # Maximum number of shared buffers
> NUMAIOVPS 1 # Number of IO vps

Are you using KIO? (that's what this would seem to indicate) check queue
length for your AIO processor. If long, you might want to add to this.

> PHYSBUFF 512 # Physical log buffer size (Kbytes)
> LOGBUFF 512 # Logical log buffer size (Kbytes)

Again, minimal effect - check the usage of your buffers.

> LOGSMAX 100 # Maximum number of logical log files
> CLEANERS 64 # Number of buffer cleaner processes

That's pretty hefty. Seems larger than would be required given the small
size of your memory footprint.

> SHMBASE 0x0 # Shared memory base address
> SHMVIRTSIZE 393216 # initial virtual shared memory segment
size
> SHMADD 131072 # Size of new shared memory segments
> (Kbytes)

Check your memory usage - are you adding a lot of memory segments? If so,
resize your SHMVIRT to try to get a single chunk of memory instead of a lot
of little ones.

> SHMTOTAL 0 # Total shared memory (Kbytes).
0=>unlimited

If you are sharing the box, then you might not want unlimited memory going
to the instance.

> CKPTINTVL 300 # Check point interval (in sec)
> LRUS 64 # Number of LRU queues
> LRU_MAX_DIRTY 10 # LRU percent dirty begin cleaning limit
> LRU_MIN_DIRTY 5 # LRU percent dirty end cleaning limit

Check how many foreground writes you are doing to tune this.

> LTXHWM 50 # Long transaction high water mark
> percentage
> LTXEHWM 60 # Long transaction high water mark
> (exclusive)

Look like the default values. Generally nothing wrong with that.

[...]

> # Backup/Restore variables
> BAR_ACT_LOG /usr/informix/log/bar_act_nazca.log
> BAR_MAX_BACKUP 0
> BAR_RETRY 1
> BAR_NB_XPORT_COUNT 10
> BAR_XFER_BUF_SIZE 31
> # BAR_DEBUG 9
> BAR_DEBUG_LOG /usr/informix/log/bar_debug.log

Ah. good.

[...]

> # Read Ahead Variables
> RA_PAGES 32 # Number of pages to attempt to read ahead
> RA_THRESHOLD 30 # Number of pages left before next group

Check how much of your read ahead you are using. If you have a lot of
users, this may be too high. For DSS this is too low.

[...]

> DBSPACETEMP temporal1,temporl2 # Default temp dbspaces

I'd have gone for 8 of these with 4 CPUs. How is temp performing?

[...]

> # Parallel Database Queries (pdq)
> MAX_PDQPRIORITY 80 # Maximum allowed pdqpriority
> DS_MAX_QUERIES 20 # Maximum number of decision support
queries
> DS_TOTAL_MEMORY 102400 # Decision support memory (Kbytes)
> DS_MAX_SCANS 1048576 # Maximum number of decision support scans
> DATASKIP off # List of dbspaces to skip

100 MB for DSS stuff. Won't get you very far - how is it performing?

[deletia]

Messages In This Thread

• ONCONFIG comments
  Bono Vox -- Tuesday, 10 May 2005, at 3:00 a.m.
  • Re: ONCONFIG comments
    Jack Parker -- Tuesday, 10 May 2005, at 6:34 a.m.
  • Re: ONCONFIG comments
    Jack Parker -- Wednesday, 11 May 2005, at 5:35 a.m.
  • Re: ONCONFIG comments
    Jack Parker -- Wednesday, 11 May 2005, at 5:55 a.m.

IDS Forum is maintained by Administrator with WebBBS 5.12.