Re: [Xosview-users] working on xosview again...

[Reed R. - r. <Rus...@ac...>]

Have you seen the 'iotop' tool?  It's similar to the third section of your tool.

" how much do you think older version need support for newer versions of xosview?

I.e. if they are upgrading it, might they not be getting new kernels?"

I know in our case, we're running RHEL5 and RHEL6 due to company standards, but we'll compile and use the latest xosview.  However, we haven't been using xosview nearly as much the past few years as we used to, so my opinion doesn't matter too much here.

-----Original Message-----
From: Linda A. Walsh [mailto:xo...@tl...] 
Sent: Monday, April 14, 2014 8:31 PM
To: Reed Russell - rreed
Cc: xos...@li...; xos...@li...
Subject: Re: [Xosview-users] working on xosview again...

Reed Russell - rreed wrote:
> I'm guessing that falling back to /proc/stat on the Diskmeter was for really old kernels.  I'm assuming that's where the disk stats were long ago.
----
I haven't seen them in /proc/stat for as long as I can remember and I was using linux in 2.0, but my memory might not be that great either!

I get disk out of /proc/partitions -- it depends on if you only want one or two devices, or all in one swoop.

And when I say I get, I mean for my other util -- a curses bases io/util that shows top net/partition and process stats that I cooked up in perl no less.

Nothing seemed to give the io stuff I wanted -- per/partition, top procs and network in a clean view...  Doesn't do anything else.  Just the io -- since nothing else did that and I didn't want to bog down refreshes doing unneeded stuff.


So it looks a bit like this...

    Mon, Apr 14  18:08:48  Period: 5s   Load:0.10   Tasks: 509   Cpu:  7.4%
――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
NetIF:◄━━━━━━━━━━━━━Receive & Transmit 2MB/s Scale ━━━━━━━━━━━━━(Tot: 
1.3MB/s)━►
eth2 :█████████████(999KB/s 
Trx)███████████                               1MB/s
eth4 :██████████                                                        
274KB/s
eth3 
:                                                                    64B/s
――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
Device      :◄━━━━━━━━━━━━━Read & Write 1MB/s Scale ━━━━━━━━━━━━(Tot: 
830KB/s)━►
Media       :█████████████████████(829KB/s Rd)█████████████████████     
829KB/s
Var         :                                                            
818B/s
Squid_Cache :                                                            
818B/s
――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
Name     :PID  :◄━━━━━━━━━━━Read & Write 1MB/s Scale ━━━━━━━━━━━(Tot: 
831KB/s)━►
transmiss:11414:████████████(562KB/s Rd)███████████                     
562KB/s
transmiss:11474:███(266KB/s Rd)██                                       
266KB/s
squid    :2691 :                                                         
818B/s
imap     :47804:                                                         
818B/s
Terminal :72186:                                                         
818B/s

Of course have to hit /proc/numbers for the individual process top users and I parse /proc/net/dev for its i/o.  Still, only about 1% cpu every
5 seconds (assuming I use default 5second period).  Too fast and it doesn't show averages as well .  Can't see it on the above, but it shows R/W in separate colors.  I run it right underneath xosview in a short Terminal window.  Down side of showing the processes is it needs root to read them all, but doesn't bother me.  Oh, and that's a real 1% of 1 cpu, not
1 percent of a 100CPU machine.  *cough*.  I hate that obscuring.  When I added the single cpu display to xosview it shows cputime/realtime as a percent, so if you hae 12 cpu's , that's 1200% of the work.  The graphs look pretty much the same... Might tinker with some logarithm scales -- It's hard to show *activity* in such a small space, that a cycling graph like for 'load', or something might be more appropriate for other graphs too since they don't have legends.

I'm still doing my work in the 1.8.4 branch -- BUT have pulled in most of mark's changes from 1.16, I'll probably bump the versio up to something higher than 16, but 1.8.4.0 was when I got the source, and never released my work before mark took over.

I bid him welcome to it at the time, as I didn't have the time to do much on it, and he wanted, to, so *dump*... sure... you can have that... ;-)

But now he's not around, and I've tried to send hime a few probs.. like the new default colors are pretty hard to see.

Also liked the wait/+io separated at the top.. away from normal cpu time (sys/user at bottom idle in middle).

how much do you think older version need support for newer versions of xosview?

I.e. if they are upgrading it, might they not be getting new kernels?

Going to try to move away from kernel versioning if I get to that and just go off how many fields are present -- more accurate and I hope the code will look better.  My c++ is very rusty and not up-to-date, so that's a minor impediment.  But oh well, way too much perl on the brain lately, I keep finding myself writing sub ....then catch myself as I start to write a type... ;-)





***************************************************************************
The information contained in this communication is confidential, is
intended only for the use of the recipient named above, and may be legally
privileged.

If the reader of this message is not the intended recipient, you are
hereby notified that any dissemination, distribution or copying of this
communication is strictly prohibited.

If you have received this communication in error, please resend this
communication to the sender and delete the original message or any copy
of it from your computer system.

Thank You.
****************************************************************************