Operating System - HP-UX
1748162 Members
3762 Online
108758 Solutions
New Discussion

Re: gunzip process stalling after every 1 min, rx2800 i4 on 11.31

 
chindi
Respected Contributor

Gunzip process stalling after every 1 min , rx2800 i4 in hpux 11iv3

Hi ,

 

Gunzip process stalling after every 1 min , rx2800 i4  , 16c / 256 gb ram , in hpux 11iv3 .

PFA kernel parms.

 

 

7 REPLIES 7
Patrick Wallek
Honored Contributor

Re: Gunzip process stalling after every 1 min , rx2800 i4 in hpux 11iv3

How do you know it stalls?  What leads you to that conclusion?

 

 

chindi
Respected Contributor

Re: Gunzip process stalling after every 1 min , rx2800 i4 in hpux 11iv3

Hi ,

 

We run a sleep of 5 sec to check the progress.

It stalls for almost after every 45 sec .

Steven Schweda
Honored Contributor

Re: Gunzip process stalling after every 1 min , rx2800 i4 in hpux 11iv3

 
Dennis Handly
Acclaimed Contributor

Re: gunzip process stalling after every 1 min, rx2800 i4 on 11.31

>to check the progress.

 

What are you checking?  The size of the output file?

How big are the compressed vs uncompressed files?

 

Most importantly, why are you even bothering checking?

All of a sudden the total time is much longer?  What changed?

 

Is this related to your other topic?

http://h30499.www3.hp.com/t5/System-Administration/wio-consistently-above-40-during-peak-load/m-p/6508662

chindi
Respected Contributor

Re: gunzip process stalling after every 1 min, rx2800 i4 on 11.31

Hi Dennis,

Its an oracle dump.gz file of 11g of size 6 gb.

We recently have upgraded to 11g from oracle 9i and facing tremendous performnce issues since .

Yes this is related to my next post as u rightly pointed it out .

 

We migrated to 11g in 3 steps ;

1] 9i to 10g migration was odne in 11iv2 server rx6600 8c/64 gb ram

2]10g to 11g migration was carried out on 11iv3 server rx2800 i4 16c/256 gb RAM .

 

In 11iv2 there were no issues in gunzipping same file .

But now the time has drastically increased by 50% aprox on these brand new v3 servers.

 

 

 

Bill Hassell
Honored Contributor

Re: gunzip process stalling after every 1 min, rx2800 i4 on 11.31

To properly compare the two servers, you'll need to provide sar -d listings from both servers  running gunzip to see if disk I/O is the same. Also, you need to see if gunzip is consuming 100% of one CPU both in the old server and now the new server. And missing from the description is what kind of disks are being used? Are you comparing internal disks on both systems? If so, how are are the internal disks setup, that is, are they part of an internal array controller? If the disks are external, factors that can affect speed are multipathing software on the 11.23 system and built-in multipathing for 11.31.

 

If you have loaded extra multipathing software on 11.31, you should disable it and configure the built-in capability for 11.31. Also, are the fibre speeds the same for external disks? You may want to run dd tests on the old and new systems to see if serial reads are similar. If you are using a new external array, has it been properly configured for HP-UX active-active support?



Bill Hassell, sysadmin
chindi
Respected Contributor

Re: gunzip process stalling after every 1 min, rx2800 i4 on 11.31

Hi Bill,

 

HPux 11iv2 disk is on EMC VNX5500 .

Dual Controller

24GB Cache

12 nos. of 8Gbps FC port

6Gbps SAS back-end architecture.

85 nos. of 300GB 15K SAS drives

58 nos. of 600GB 15K SAS drives

8 nos. of 2TB NL SAS drives

5 nos. of 100GB SSD

Yes we have loaded EMC multipathing software on it .

 

 

HPUX 11iv3 is on HP 3PAR.

No multipathing software on it.