query on export node

[storageadmin]

Hi all,

I am working on TSM 5.5 on RHEL 4.5 with quantum scalar i500 library having 4 drives (Gen4).

My database was 80Gb and was max utilized. Server as become very slow. Expiration was not run since long days. so we started expiration..now it as come to 30% and also 50% of tapes where reclaimed. but even then my backup or archieve speed is avg 1-2 GB per minute with 4 drives. Generally it should be 4+ GB per minute.

some archive sets are fine but few archieve sets take 10 mins time to open. They are no network or servers issues also.

do suggest what could be done to improve the server performance.


Regards,
Storageadmin

 

[storageadmin]

server performance is poor..suggest

Hi all,

I am working on TSM 5.5 on RHEL 4.5 with quantum scalar i500 library having 4 drives (Gen4).

My database was 80Gb and was max utilized. Server as become very slow. Expiration was not run since long days. so we started expiration..now it as come to 30% and also 50% of tapes where reclaimed. but even then my backup or archieve speed is avg 1-2 GB per minute with 4 drives. Generally it should be 4+ GB per minute.

some archive sets are fine but few archieve sets take 10 mins time to open. They are no network or servers issues also.

do suggest what could be done to improve the server performance.


Regards,
Storageadmin

sorry i have given wrong heading as query in export node above

 

[Efim]

Try to run server instrumentation for find bottleneck.
When the system show slow performance run dsmadmc
type:
INSTrumentation Begin
wait 10-20 min
type
INSTrumentation End > "path to file"

The file will be not huge and show this file to us. I think it is the simplest way.
Efim

 

[storageadmin]

Instrumention report

Hi Efim,

Please find the attached Report ..

i have run this command when no process was running..server was idle.

do suggest

Regards,
storageadmin.

 

[Efim]

Sorry, i miss you replay.

Run INSTRUMENTATION during server activity (backup node). In this case you can see witch treads are waiting and witch use more time than normal.

The second step will be run instrumentation on client side for determine bottleneck.
Without this i see that Network Recv spent 913.737 while other Threads are waiting, but you told that server is idle and i dont know what server did with network...

Efim

 

[storageadmin]

Hi Efim,

Thank u for u'r suggestion. I am attaching new report which was taken when archive was running.

I will soon send the client report also.

do suggest..


Regards,
Storageadmin

 

[storageadmin]

Hi Efim,


Thank u for u'r suggestion. Before 10 days i have run audit DB cmd with fix=yes option. After completing that..for 10 days my backup's..archieve were running fine.

But since 2 days backup's..are runinng fine but archive have again become slow..and if i retrive those archives..it takes 5-10 mins of time to open.


I am attaching new report which was taken when archive was running.

I will soon send the client report also.

do suggest..


Regards,
Storageadmin

 

[storageadmin]

performance

Hi Efim,

Yes that report was not complete. I am attaching new Instrumentation report that was taken today when archive was running.

also find the attached dsmerror.log report.

please suggest what this below errors are that was in dsmerror.log

1) truncated while reading in Shared Static mode.


Regards,
Storageadmin

 

[storageadmin]

performance

Hi Efim,

I have one more issue to clarify. If TSM has I/O problem..it should even effect the backup performance also. But backups are running fine.

I am attaching instrumentation reports taken when backup was running, archives running from FC/Sata volumes and also client instrumentation report.

please clarify..

Regards,
Storageadmin

 

[Efim]

You make a lot of traces....
ok
1. You use TSM client v.5.3.0.0 with TSM 5.5 server.
2. Good performance:
fc_inst_rep_23june08.txt
inst_bkp_rep_24jun08.txt
3. Bad performance
sata_rep2_inst_23jun08.txt

4. dsminstr.report.p13387.txt - it is not full report.....

I see that you use different quantuiy of objects when you backup and archive. And we can't compare peformance.
The size of you database is 80Gb whith is maximum recommended for best pepformance.

As i understend sata_rep2_inst_23jun08 - ARCHIVE activity.
I can't explain why when you do archive database take half of ARCHIVE time.

Efim