Linux Free Command (Analysing RAM utilization)

I was reading these and thought sharing them will be a useful.

The linux free command allows us to check free/used memory on the system The output below is the result of running free -m on my system (-m means output is in MB):

total used free shared buffers cached
Mem: 3856 1121 2735 0 17 180
-/+ buffers/cache: 923 2933
Swap: 2533 1044 1489

[edit]
Output explained

The first line starting with Mem: gives us the following information:

* total - is the total avaialble RAM (== Physical Memory) after subtracting the amount used by the kernel! In my case I have 4GB RAM and the total displays less than this.
* used - is the part of the RAM that currently has information in it that can be of use to the system (remember: unused RAM is useless, try to maximise this value)
* free - is just total-used
* shared - is the amount of memory shared between processes
* buffers and cached - the cached data and buffers for IO

The second line starting with -/+ buffers/cache: tells us how much of the memory in the buffers/cache is used by the applications and how much is free. Keep in mind that in general the cache is filled with disk IO cached data. The cache can be very easily reclaimed by the OS for applications. Let BUFFERS + CACHED from first line be value X.

X subtracted from the USED memory from the first line gives how much RAM is used by applications (USED value on second line)

X added to the FREE memory on the first line gives how much RAM applications can still request from the OS.

While the first line handles the values of currently used RAM, including applications and caches (but excluding kernel), the second line gives info on application related memory: how much is currently in use and how much is there still available for the applications.

You can find more info here

Using Find output and making a grep on it

Some times, we may be in situation to find the list of files in a recursive directory and search for a specific contents in it.

Below is the command which does that. In below command I'm trying to find commons-beanutils.jar recursive in a directory and a grep command on the listed files to actually find do the jar files has class file 'BeanUtilsBean.class in it.

find . -name *beanutils*.jar -exec grep 'BeanUtilsBean.class' '{}' \;

How to view and change the timestamp of a file say creation date

'stat' command will give the last-accessed/last-modified/file-created date with timestamps in it.

[prakashc@cu120 jbossweb-tomcat50.sar]$ stat commons-beanutils-1.8.0.jar
File: `commons-beanutils-1.8.0.jar'
Size: 231320 Blocks: 464 IO Block: 4096 regular file
Device: 303h/771d Inode: 6547316 Links: 1
Access: (0700/-rwx------) Uid: ( 7069/prakashc) Gid: ( 4001/__cubitu)
Access: 2011-02-09 07:01:55.000000000 +0530
Modify: 2011-02-09 05:50:40.000000000 +0530
Change: 2011-02-09 06:12:08.000000000 +0530

using 'touch' command you can modify the date created

touch commons-beanutils-1.8.0.jar -t [[CC]YY]MMDDhhmm [.SS]

MM - The month of the year [01-12].
DD - The day of the month [01-31].
hh - The hour of the day [00-23].
mm - The minute of the hour [00-59].
CC - The first two digits of the year.
YY - The second two digits of the year.
SS - The second of the minute [00-61].

Remove duplicate rows from postgress using ctid

Today, I was in a situation to delete the duplicate row entries in a table which has all the column values similar including the unique constraints too.

I'm not sure how the table accepted the unique constraints of same value, it's been an large dataset and I guess some could've gone wrong on migration or some thing else.

While trying to reindex the table it failed stating

>>

reindexdb: reindexing of database "testdb" failed: ERROR: could not create unique index "folder_pk"

DETAIL: Table contains duplicated values.

<<


After trying to remove the duplicate values, I found all the columns values are similar including the primary keys,

So finally got an idea from one forum, which suggests to use the "ctid" of the rows to delete that.

It looks, for the table rows, internally there is ctid and you can view it using

select ctid, id from table;

Then, remove the specific rows which you need using the ctid.

"Lockness" eclipse thread dump analysis plugin

There is a eclipse plugin "Lockness" for thread dump analysis (http://lockness.plugin.free.fr/setup.php). Using this we can analyze the thread locks to diagonalize the problems. To download (http://lockness.plugin.free.fr/archive/com.cr.lockness.Lockness_3.0.0.jar), put this jar to eclipse3.x/plugins dir and restart eclipse.

How to take a thread dump:

Sending "kill -3 signal to the JBOSS/TOMCAT process" will give you the whole thread dump. Ex: kill -3 #### -> process id.

After executing the above command, you can find thread dump in catalina.out for tomcat or server.log for jboss.

You've to trim the thread dump contents exactly from above log files, such that the dump file starts with "Full thread dump" and end with "VM Thread".

Also, if you have some timestamps added in thread dump, you can trim that using below cut command. For ex: say each line has some time stamps added like "2010-10-25 13:29:23 PDT", then to skip the first 3 columns. Use below command

cat threaddump.txt | cut -d " " -f4- > cut_file_threaddump.txt

The above command will skip first 3 columns and saves rest of the columns what we needed to cut_file_threaddump.txt.

Then from eclipse, create a new java project, create folder pointing to above create dump dir and opening this using "Lockness" thread dump analyser plugin will give you the analysis.