LinuxQuestions.org
Review your favorite Linux distribution.
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - General
User Name
Password
Linux - General This Linux forum is for general Linux questions and discussion.
If it is Linux Related and doesn't seem to fit in any other forum then this is the place.

Notices


Reply
  Search this Thread
Old 12-09-2011, 02:21 PM   #46
markseger
Member
 
Registered: Jul 2003
Posts: 244

Rep: Reputation: 26

I guess the one last piece that's missing is seeing what the system is doing when not copy a file but just creating one, but using a tool like dt or dd. Remember, if you're reading and writing from/to 2 different locations on the same disk you're not going to be as efficient. As an example, here's what I see on my system with collectl when I run dt one window:

Quote:
[root@poker collectl]# ./dt of=/tmp/test limit=5g bs=1m disable=compare,verify
and collectl in another:

Quote:
[root@poker collectl]# collectl -scdm
waiting for 1 second sample...
#<--------CPU--------><-----------Memory-----------><----------Disks----------->
#cpu sys inter ctxsw Free Buff Cach Inac Slab Map KBRead Reads KBWrit Writes
0 0 1102 274 4G 228M 1G 1G 186M 764M 0 0 0 0
10 10 1126 330 4G 228M 1G 1G 189M 765M 0 0 0 0
25 25 1103 295 4G 228M 1G 1G 204M 765M 0 0 0 0
25 25 1121 300 3G 228M 2G 2G 219M 765M 0 0 0 0
28 27 1180 337 3G 229M 2G 2G 239M 765M 0 0 27084 78
26 26 1209 338 3G 229M 2G 2G 255M 765M 0 0 39596 88
26 26 1178 317 3G 229M 2G 3G 270M 765M 0 0 38400 77
26 26 1197 328 2G 229M 3G 3G 285M 765M 0 0 38362 75
26 26 1178 302 2G 230M 3G 3G 300M 765M 0 0 38400 75
26 26 1199 328 2G 230M 3G 3G 315M 765M 0 0 38438 76
26 26 1179 306 1G 230M 4G 4G 330M 765M 0 0 38400 75
26 26 1196 349 1G 231M 4G 4G 344M 765M 0 0 38951 76
you can actually see the cache start to fill for several seconds before any disk i/o occurs. then the writing starts to disk and it goes very smoothly, chewing up about 25% of the cpu. Actually when I run with -sC in another window I can see at one point it's two cpus loaded at 75/25% respectively and at another point it's 100% of one CPU.

Quote:
# SINGLE CPU STATISTICS
# Cpu User Nice Sys Wait IRQ Soft Steal Idle
0 0 0 74 0 0 0 0 26
1 0 0 25 0 0 0 0 73
2 0 0 3 94 0 2 0 0
3 0 0 0 100 0 0 0 0
0 0 0 99 0 0 0 0 0
1 0 0 0 0 0 0 0 100
2 0 0 1 94 0 2 0 0
3 0 0 0 99 0 0 0 0
0 0 0 100 0 0 0 0 0
1 0 0 0 0 0 0 0 100
2 0 0 3 93 0 2 0 0
3 0 0 0 100 0 0 0 0
This is the level of detail you need to get into if you really want to see what's going on.

-mark
 
Old 12-09-2011, 04:56 PM   #47
markseger
Member
 
Registered: Jul 2003
Posts: 244

Rep: Reputation: 26
having nothing to do with this specific problem, I'm concerned over the errors from running "collectl --top rss" posted earlier as:

Quote:
Use of uninitialized value in division (/) at /usr/share/collectl/formatit.ph line 6981.
Use of uninitialized value in division (/) at /usr/share/collectl/formatit.ph line 6981.
974 root 20 958 0 S 39M 24M 0 0.06 0.13 19 11:30.02 0 0 1 1839 /usr/bin/X
there were other errors too, but rest assured both those have already been identified and fixed. But I'm really stumped over these. From what I can tell they're referring to the lines:

Quote:
$line.=sprintf("%4s %4s ",
cvt($procRKB[$i]/$interval2Secs,4,0,1),
cvt($procWKB[$i]/$interval2Secs,4,0,1)) if $processIOFlag;
which calculate the process I/O stats. So either the $proc vars are not set OR $interval2Secs is not set. The only thing is that second variable is used elsewhere which makes me think the problem is with the $proc variables. But not having access the system on which this is occurring I can't verify anything.

can anyone else reproduce this problem? if so, can you see what's in /proc/pid/io, where 'pid' is the pid of the process generating the errors? is it possible this is defined for some but not all processes? I'd really like to get to the bottom of this as this the sort of thing that could give collectl a bad rep and clearly this is pretty rare as I've certainly never seen it nor has anyone reported it. could it be something about this specific release of ubuntu?

If someone does want to dig deeper and you identify a specific process for which this is failing, you can always try the command

Quote:
collectl -i:1 -sZ --procfilt cpname -c1
where pname is the process name, the leading 'c' is required. it will display a single line of output and exit

if this line fails as well, tack -d4 onto the end. that puts collectl into a special debug mode in which the contents of /proc are displayed and maybe there will be a clue there.

thanks in advance

-mark
 
  


Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



Similar Threads
Thread Thread Starter Forum Replies Last Post
Transmission torrent client takes up too much resources at fast downloading. MheAd Linux - Software 15 05-02-2010 05:49 PM
core process eating up resources jaggy00 Linux - Newbie 2 04-23-2007 05:11 AM
copying to my usb2.0 disk makes system realy slow and takes up to 90% cpu Sammael Linux - Hardware 0 08-21-2006 06:18 AM
copying from pc to usb-connected mp3-player takes ages; suse 10.0 ungua SUSE / openSUSE 2 03-28-2006 07:00 AM
Sendmail takes too mush system resources executor Linux - Software 0 12-14-2005 06:23 AM

LinuxQuestions.org > Forums > Linux Forums > Linux - General

All times are GMT -5. The time now is 07:08 PM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
Facebook: linuxquestions Google+: linuxquestions
Open Source Consulting | Domain Registration