I have read many articles on hdparam to calculate the disk read and write speeds and some on interface and CPU limits. But is there a structured way of calculating the maximum throughput of a server including all the subsystems. Like storage, CPU, network, memory and so on? So that I can create a script that i can run on a newly installed Linux machine and calculate the maximum throughput .
View 4 RepliesHow do I enable buffered I/O for increased disk throughput on Linux ?
View 3 Replies View RelatedI have an enormous quadcore machine with 16gb ram and dual gigabit NICs. It used to be for MySQL but we have upgraded the whole database infrastructure so now this server is left floating. I had the great idea of turning this into a reverse-proxy (using apache mod_proxy) and it really handles a ton of requests. But I have a feeling that we are not getting the most use out of what it can offer.
Our traffic consists of a few thousand very small (less than 10 byte) ajax calls per second, and frequently I find we are running out of kernel allocated network stack to handle all the requests. Often we get the kern.log warning "possible SYN flooding on port 80. Sending cookies." and other things like this. Obviously we are not getting SYN flooded, we just have very high demand.
So far I have found a few kernel tuning guides to tell the kernel to allocate more of the base system memory for networking but every guide I have found has been for the purpose of increasing the performance between WAN links (direct backbones between offices etc) and usually with very large file sizes being the priority. One such example (and great) write up is here:
cyberciti.biz/faq/linux-tcp-tuning/
I was hoping some people could provide further input, such as along the lines of disabling nf_conntrack (to speed up socket set up/tear down time) or anything that will speed up a high throughput proxy like mine. Any links to studies or benchmarks between different configurations or hardware gets extra points!
I have a test RHEL5 box, sitting in a brand new Dell blade rack on a PowerEdge M610, with a lovely Emulex OCm10102-F-M FCoE card connected to a Cisco Nexus 5000 switch. The whole setup is extremely new (the cards only recently became available for purchase.) We've finally worked with Emulex to get the cards functional, and we are ready to some stress testing to the SAN. My question now is, is there a good tool I can use to generate a large amount of traffic to a LUN? The Wintel team used a windows-only tool that showed an average of 6gigs/second throughput, so I need something that can generate very large files and simulate a consistent throughput to the LUN. I found iozone, but I'm having a devil of a time with it.
View 2 Replies View RelatedMy monitor tools indicate that the percentage of swap used is more than the ideal , so I want to calculate which process is consuming from the swap. I tried the following but it didn't help be 100%
#ps -eo vsz,rss,pid,args | sort -n
How to accurately calculate which process consuming from swap, just the swap no ram + swap!
I am writing my own syslog collector, but im stuck with calculating the severity of the message. The PRI value is calculated:-
Code:
(Facility * 8) + Severity
But from the context of a collector, neither Facility or severity would be known? You need to know either severity or Facility to calculate facility or severity?! The RFC goes in to no more detail about calculating the severity value, other tan this expression.
I am trying to update from ubuntu 10.04.1 to 10.10 using the do-release-upgrade command. I get this error:
Code:
An unresolvable problem occurred while calculating the upgrade: The package 'update-manager-core' is marked for removal but it is in the removal blacklist.
This can be caused by:
* Upgrading to a pre-release version of Ubuntu
* Running the current pre-release version of Ubuntu
* Unofficial software packages not provided by Ubuntu
The first two can't be correct right? So... If the third one is true. How can I find these packages? Or, is there any other way to upgrade to 10.10?
I have implemented level 5 software raid (of 4 partitions on a SINGLE external hard disk). Implementing it on a single hard disk doesn't serve much purpose, I understand, but I did it for demonstration purpose only. The problem is when I read from the raid device into my laptop's hard disk, the time taken to transfer a 900 MB file is about 4 minutes (4 GB RAM) whereas the rime taken to transfer it from a pen drive is 43 seconds. Isn't raid supposed to work faster. However, the write to the raid device is faster than to a pendrive. Shouldn't it be taking data in parallel from the 3 partitions (as even a single disk has multiple heads) ?
View 2 Replies View Relatedi want to know the throughput of my pc after i enable RFS and RPS configuration. how to calculate tps or how to see throughput ......
View 1 Replies View RelatedMy objective is when the throughput is increase my shell script is Print a message. But the Question is how to check the bandwidth.
View 1 Replies View RelatedI want to know the actual maximum throughput of my lan at home.Could you suggest some free monitoring tools for me to use
View 2 Replies View RelatedLets assume we have a 100mbps connection between two servers in same switch. then How much is the total throughput of the connection between servers. For example if I am downloading with 70mbps from other server , do I still have 100mbps upload BW available ? or 30mbps is available ? in other words for a 100mbps connection , the limit of download+upload is 100mbps ? or we have 100mbps download and 100mbps upload available ?
View 12 Replies View RelatedI have pc with two NIC's (eht0 and eth1). I want to test throughput switch/router.I can ping on both interfaces.Connection Diagram:
eth0 -> Patchcord -> testing devices -> Patchcord -> eth1
I run two iperf:
-s -B 192.168.1.10 (eth0)
-c 192.168.1.20 -B 192.168.1.20 (eth1)
Results:
[ 3] 0.0-10.0 sec 4.45 GBytes 3.82 Gbits/sec
The test results are false because I use 100Mb/s NIC's. I can't use two pc's for test.I tried to do the same test under windows 7 x64 it and works properly, but under windows xp there occurs the same problem like under ubuntu.
I accepted the latest Natty update about which I was notified a couple of days ago (I don't remember exactly). I don't remember a new kernel being pushed out, but for the record I'm on 2.6.38-8-generic right now. I'm trying to interact with a USB compact flash reader and I'm getting ridiculously low throughput (less than 1 MB per second). Before this latest upgrade, I don't remember this being a problem.
Better yet, does anybody know what might be wrong and what I might be able to do to fix this?
Having finally got Samba shares and printer shares working on my Lucid Lynx server, I've noticed that the throughput from/to those shares is very slow to what I was used to when I had those shares on a W2K server.Having Googled this problem, I noted that there was a known issue with this in Karmic (believe Brian Wu was looking at it). However, I don't see any results in Google as to this being a problem in Lucid.
I know ipv6 can cause a problem in this regard but having issued the command:
lsmod | grep inet6
I know that ipv6 is indeed disabled on my Lucid server. Any further information on whether the error reported in Karmic still applies in Lucid, and can this be addressed in any way?
calculating the IOPS for my system ?I had understood that IOPS could be calculated by monitoring IOSTAT. The r/s (reads/sec) and w/s (write/sec) combined basically gave you the peak IOPS.However having monitored iostat for my app during a performance test, the figures I'm seeing lead me to believe that iostat couldn't be a reliable way to calculate IOPS. I get values of over 10,000 for w/s! I didn't think this was possible given the storage system i have (4 x 7500rpm sata disks in raid 5).
View 1 Replies View RelatedWith array in php. All I was having the following columns:
Code:
Now Like the Above there were more variables defined through script:
Code:
Now what the above code is doing is calculating average of above values from columns{1-?,4-?)..likewise
Say If I remove two Machine and calculate the avg.
Recently I was trying to make a quick script for calculating payment between workers. I got stuck however. Here is what I am in for...
#!/bin/bash
#Setting the calculator
calc=` bc << EOF
scale=4
a=$2*($3/100)
b=$4
a/b
EOF
`
I've saved the $1 for further script.
#Creating the script
while [ -n $1 ]
do
case $1 in
workerA) read -p "Enter the total sum: " total
total=$2;
read -p "Enter the percentage:" percent
percent=$3;
read -p "Enter number of workers:" workers
workers=$4 ;;
*) echo "Error usage!"
break
esac
done
My script ends here. It just does not work. Did I sat the vars in the wrong way?
What originally seemed like an easy thing to calculate has given me a big headache. Perhaps someone can help me with my issue. I am trying to find, in particular, how much memory certain application processes are taking. The process always is the same name, main_server, but with an argument to tell it what to do when running as a daemon.
When running the following command against all "main_server" processes, it produces a result in megabytes based on the output of the rss field in 'ps'.
Code:
CALC=0
for ea in
`ps -e -orss=,args= |
sort -b -k1,1n |
pr -TW$COLUMNS |
grep main_server |
grep -v "grep main_server" |
[Code]...
Currently I am left scratching my head. For capacity planning purposes, it would be nice to know how many more 'main_server' processes could run on the system without causing it to swap. Knowing the buffer and cache usage will go down as running processes demand more memory, I prefer to look at the free memory excluding cache and buffers. However, since 'ps' is reporting the processes are using more memory than free reports is in use without those things, I have no way to know how many more processes the system can support. I played around with different fields in 'ps', such as vsize, size, etc, but with no luck in matching up any numbers.
I'm trying to create a script that will find the bandwidth usage of certain protocols only. For example, SMTP. I would like it to just return a number. Is there a known command/parameters to output something like this?
View 2 Replies View RelatedI'm trying to write a python script that will calculate my test scores from a text file named "exams."
Sample input would be:
Output will look like:
I have the following code:
Code:
My intent is to sum the values found in L and divide by len(L).
I think my issue is that when I split the values, I get a string array (ex. ['85', '90'] ), which makes it hard to do calculations.
I'm writing a script which will calculate bandwidth usage, I'm monitoring the bandwidth with cacti and the results don't quite line up, can someone either correct my script (if it contains faults) or tell me why the results aren't the same. I'm using a guide obtained from cisco here: [URL]. Here's my code:
Code:
SNMPCOMMUNITY=public
IP=127.0.0.1
INTNUMBER=1
OUT=$(snmpget -v2c -c $SNMPCOMMUNITY $IP ifOutOctets.$INTNUMBER | awk '{print $4}')
[Code]....
When I run df it shows the root device is full.
Filesystem Size Used Avail Use% Mounted on
/dev/sda1 9.9G 9.4G 0 100% /
I looked at the inode usage and there is pretty much space available for root device
Filesystem Inodes IUsed IFree IUse% Mounted on
/dev/sda1 640K 103K 538K 16% /
But, when I run the du command, it shows I have used only 2G out of 9.9G.
ip-10-204-70-44:/$ du -xh --max-depth=1
14M ./etc
4.0K ./mnt[code]....
It just driving me crazy and interesting too. This is big problem for us since the root disk / is full and some of the function in our site is failing.
I have a data in a column.
Code:
I want to calculate average and standard deviation. As first step I want to calculate average for the data than calculate (del=data - avg) for all the data.
I suppose get
Code:
For this I use AWK and the code goes like this
Code:
But I get different answers.
Code:
Why the answers are so different? since this is wrong I can not continue calculating the standard deviation.
i have installed ns 2.33 and added the patch mobiwan for supprting mobile ipv6. i ran some tcl scripts , and below is the trace file generated. calculate the handoff latency, and if someone has awk or perl script to do so, tcl file [URL] out.tr [URL]
View 1 Replies View RelatedActually i m new to scripting in Linux, we have RedHat Server Edition 4 in which different software licences are installed and software itself installed at client side. the usage of the software is recorded daily in a text file which is of format
I want to write a script that checks that for any OUT & IN, if utility, user and Client PC Name are same then subtract OUT Time from IN Time. This has to be done for each & every user and utility.Is there any GUI programming possible in Linux.
I have just tried to upgrade to Ubuntu 10.10 from 10.04 using the update manager
It asks me for me password and then a window opens saying "Upgrading Ubuntu to version 10.10"
It gets to "setting new software channels" then a message pops up saying "third party sources disabled" next a little error icon appears in the Notification Area then another message saying: Could not calculate the upgrade
An unresolvable problem occurred while calculating the upgrade: E:Error, pkgProblemResolver::Resolve generated breaks, this may be caused by held packages.
This can be caused by:
-Upgrading to a pre-release version of Ubuntu
-Running the current pre-release version of Ubuntu
-Unofficial software packages not provided by Ubuntu
If none of this applies, then please report this bug against the 'update-manager' package and include the files in /var/log/dist-upgrade/ in the bug report.
I am new to shell script and to this form as well, I did try to search for a similar post like mine here, but could not find one.
Here is what I'm trying to do:
I am trying to grep server logs to find a specific string and then capture the time stamp and the value of that grep string in them. The log file prints out messages on per sec basis.
My script is able to grep the server logs for the entire period of my load runs and then outputted it to a .csv file too.
Unfortunately this .csv file is too large to extract it on my PC and to generate graphs as it exceeds the excel limit. I need some help on how to read this .csv file in a shell script and then take an average on per min basis before I can export it out on my desktop and generate graphs for analysis. example of the out in my .csv file:
I have a Acer Revo R3610 nettop on which I installed oS11.3. It has an Atheros AR5001 wireless network adaptor.I'm getting very bad latency and throughput on with this card. In comparison my Thinkpad with Intel card achieves at least 10 times the throughput from the same location. Also when I boot my wifi will come up authenticate and then a few seconds later disconnect and reconnect. I've tried updating the kernel to 2.6.35 as one of the fixes was to improve the performance of the ath5k driver. I'm using wpa2 authentication.
Here are some details:
Code:
mediacentre:~ # /sbin/lspci -nnk
00:00.0 Host bridge [0600]: nVidia Corporation MCP79 Host Bridge [10de:0a82] (rev b1)
Subsystem: Acer Incorporated [ALI] Device [1025:0222]
[code]....
Is there anything else I paste that will help determine the cause of the bad throughput. Browsing the web or copying files across the network is bordering unusable.
I have an array called arrayini which stores numbers. I want to take log to the base 2 of each of the numbers in that array and put it in file called result. I've used the following code to do it.
Code:
size=${#arrayini[@]}
for ((i=0;i<size;i++))
do
echo "scale = 12; l(${arrayini[$i]})/l(2)" | bc -l
done >result
It works fine but its taking pretty long to calculate since I've got about 230,000 items in the array. So I decided to store the result into an array hoping that it'd be faster. I tried the following code. arrayresult is where I try and store the result. The code doesn't work because of the second last line.
Code:
unset arrayresult
size=${#arrayini[@]}
for ((i=0;i<size;i++))
do
arrayresult[$i]="scale = 12; l(${arrayini[$i]})/l(2)" | bc -l
done >FILE2
There is a syntax error clearly.