General
cancel
Showing results for 
Search instead for 
Did you mean: 

How to check if the load-balancing is happening in 11.31

Avinash20
Honored Contributor

How to check if the load-balancing is happening in 11.31

Do you know how to check if the devices are load-balancing when under native MPIO on 11.31.

I would appreciate if you could provide me the OS command rather than the software(Glance etc..).
Am aware of sar, iostat which doesn't serve my purpose.

For eg: In EMC powerpath, the output would be

# powermt watch dev=disk635
Pseudo name=disk635
Symmetrix ID=000XXXXXXX
Logical device ID=161C state=alive; policy=SymmOpt; priority=0; queued-IOs=204;
==============================================================================
--------------- Host ---------------         - Stor - -- I/O Path -- -- Stats ---
### HW Path                                                                 I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
0 0/3/1/0.0x50060XXXXXXXXXXX.0x4032000000000000 c7t6d2     FA 9aB active alive 105 0

1 0/4/1/0.0x50060XXXXXXXXXXX.0x4032000000000000  c11t6d2    FA 9aB active alive  99   0

Here's the QIO's will show the IO against each path..
Is there any command in Unix (scsimgr) which shows the same result.

"Light travels faster than sound. That's why some people appear bright until you hear them speak."
4 REPLIES
Earl_Crowder
Trusted Contributor

Re: How to check if the load-balancing is happening in 11.31

scsimgr is the key:

 

DISKDEV=/dev/rdisk/disk26

scsimgr get_stat -D ${DISKDEV} | grep Bytes

scsimgr lun_map -D ${DISKDEV} | grep Hard | awk '{print $4}' | while read line; do
        scsimgr get_stat -H ${line} | grep -i Bytes

done

Avinash20
Honored Contributor

Re: How to check if the load-balancing is happening in 11.31

Thanks for your reply..

This provides me with output of disk26 only.

what I am looking here is if disk26 has two paths going to two different Fibre Channel ports.

If there are high IO via one path, the native mpio should automatically load balance depending the the algorithm we are using.. So I would see IO going via both the Legacy devices..

I am under beleive that scsimgr command would require an enhancement to have this output, like

 

Agile device: /dev/disk/diskXX <No of IO going via this path at present>

Legacy device:

/dev/dsk/cAtYdZ <Path> <No of IO going via this path at present>

/dev/dsk/cBtYdZ <Path> <No of IO going via this path at present>

 

or something like

Agile device: /dev/disk/diskXX <No of IO queued>

Legacy device:

/dev/dsk/cAtYdZ <Path> <No of IO queued>

/dev/dsk/cBtYdZ <Path> <No of IO queued>

 

# ioscan -m dsf

/dev/rdisk/disk26         /dev/rdsk/c2t0d0
                         /dev/rdsk/c1t0d0

"Light travels faster than sound. That's why some people appear bright until you hear them speak."
Robert_Jewell
Honored Contributor

Re: How to check if the load-balancing is happening in 11.31

Perhaps using sar will help some.   -H reports on HBA activity, -L provides lunpath info.

 

A couple of examples:

 

# sar -H  1 1

HP-UX hprx4640 B.11.31 U ia64    08/24/11

12:12:06     ctlr   util t-put  IO/s   r/s   w/s   read  write avque avwait avserv
                   %age   MB/s   num   num   num   MB/s   MB/s   num   msec   msec
12:12:07    c8xx2    98   2.08   287   287     0   2.08   0.00     1      0      7
             fcd2    41  24.51   394   253   141  15.84   8.67     9      0      0

# sar -H -L 1 1

HP-UX hprx4640 B.11.31 U ia64    08/24/11

12:10:23     ctlr   util t-put  IO/s   r/s   w/s   read  write avque avwait avserv
                   %age   MB/s   num   num   num   MB/s   MB/s   num   msec   msec
                                  lunpath   %busy   avque     r/s     w/s  blks/s  avwait  avserv
                                             %age     num     num     num     num    msec    msec
12:10:24    c8xx2     3   0.10    11    10     1   0.10   0.00     1      0      2
             fcd2    77  85.83  1429   820   609  48.70  37.13     9      0      0
                          disk16_lunpath1    2.00    0.50      10       1     208    0.00    2.50
                         disk17_lunpath18   69.00    0.50     820       0   99736    0.00    0.85
                         disk18_lunpath19   22.00   48.76       0     608   76042   19.09    2.78
                         disk20_lunpath21    1.00    0.50       0       1       4    0.00    8.82

 

-Bob


----------------
Was this helpful? Like this post by giving me a thumbs up below!
vjta
Regular Advisor

Re: How to check if the load-balancing is happening in 11.31

scsimgr lun_map

Vijeta Bhedi