Cause of Longer Backup Time

[nibble]

hi guys, what are the possible reasons why backup of informix (ontape) would take longer that the usual time?? from 2 hrs it became 15 hrs... this 2 hours duration has been the average for 3 yrs... cud there be any configuration problem? how abt disk spaces for db...btw, its an informix on top of ux11.00

[MarkSyder]

Do you have any performance monitoring tools (e.g. Big Brother) on your system that would show if disc space or anything similar has suddenly become a problem?

Try running top to see if you have any processes hogging processor time.

Mark Syder (like the drink but spelt different)

[nibble]

thanks.. got this one

System: XXX
Wed Jan 28 18:32:24 2004
Load averages: 2.23, 2.40, 2.25
202 processes: 199 sleeping, 2 running, 1 zombie
Cpu states:
CPU LOAD USER NICE SYS IDLE BLOCK SWAIT INTR SSYS
0 2.41 0.6% 0.0% 3.0% 96.4% 0.0% 0.0% 0.0% 0.0%
1 2.06 7.4% 0.0% 8.2% 84.5% 0.0% 0.0% 0.0% 0.0%
2 2.08 4.0% 0.0% 7.0% 89.1% 0.0% 0.0% 0.0% 0.0%
3 2.39 2.8% 0.0% 4.0% 93.2% 0.0% 0.0% 0.0% 0.0%
--- ---- ----- ----- ----- ----- ----- ----- ----- -----
avg 2.23 3.8% 0.0% 5.6% 90.7% 0.0% 0.0% 0.0% 0.0%

[Steven E. Protter · Accepted solution]

Lots of possible causes.

The database could be not down properly.

If through a network there could be a configuration, switch or wiring issue.

lanadmin -x 0

change 0 to the real lan number

Could be lack of free space on the /var filesystem where some backups are temporarily spooled.

Check these things.

I'm attaching a performance monitoring script.

SEP

[nibble]

i dont find anything unusual here..what do u think?

[MarkSyder]

The output from top looks fine.

SEP has suggested checking /var - what does bdf show for that?

Mark

[nibble]

all FS are ok. they are less than 60%... probably, this could be database related and database needs to be re-started.

thanks a lot guys

[nibble]

hi SEP, one more thing... should i create the /var/adm/sa/sadd directory? its not existing... sar cant work

[Jeff Schussele]

Hi nibble,

You must give iterations & time values to sar commands - EX:

sar -d 10 5

This would give stats of 10 iterations five seconds apart.

Rgds,
Jeff

[Mic V.]

If you're using a client/server backup (I assume you are), check the backup server. I've seen situations slow backups influenced by tape drives going bad, media with errors, SCSI cards going bad, inability to stream enough data to tape, someone clogging the network with unusual data transfers...

For example, with Legato Networker, you can watch (via nwadmin or nsrwatch) the speed of a tape write. I learned what a "normal" speed was and then I could detect problems.
You can also examine logs created by the backup software (homegrown or otherwise).

[Sanjiv Sharma_1]

Hi Nibble,

I believe you are taking a local ontape backup of the Informix Database.

You can do the following.

1. Clean the tape drive with a cleaning tape.
2. Take the backup in a new media.
3. Shutown/Startup the Informix database if possible.
4. Restart the Server if possible. Worst Case.

HTH,

[nibble]

hi sanj,

ive already done cleaning the drive, using a new tape and restarting d dbase... we're running the backup now and observing. if still d same, i believe we need to restart the server..

but u guys rule out the data itself that cud grow large? but i guess its too much for an increase of 10-12 hours of backup. this shud be abt 1 or 2 hrs only if data has really grown.

any other possibilities?

[Bill Hassell]

What is the backup method. Most Informix sites run raw volumes so if the backup is simply the raw data, amount of data in the backup will be the same every time. However, if you are exporrting the database into files, write into your backup script a routine that summarizes all the file sizes before the actual backup. The total will give you a better idea of what you are backing up.

Another possibility is that the backup path from disk to tape (is it over a LAN? are there any other computers connected to the dedicated backup LAN?) Modern tape drives are streamers and waste an enormous amount repositioning when the data stream can't keep up with the tape drive. A single misconfigured PC can flood the network with useless packets and destroy the backup throughput.

[JJ_4]

What version of Informix??

When it is running do "onstat -g ath" a few times.

You want to get the stack of the arcbackup1 thread onstat -g stk

[bill summers]

Check syslog for bus hang or SCSI errors.
Could be faulty SCSI connection to DLT.