Gl.ib.ly

Lets get straight into the action:sudo seq -w 1 30 | awk '{ printf $1","; system("grep " $1 "/Nov/2008 /var/log/apache2/ssl_access_log | grep login | grep POST | wc -l | tr \"\\n\" ,"); print "" }'. This code produces CSV output in the format DAY,LOGINS,.

Lets break all this down:

• /var/log/apache2/ssl_access_log - This is my SSL access log (it is actually a merge of my November files which have been logrotated


• seq -w 1 30 - This gives us values 01 to 30. The -w bit gives us the leading zeros do man seq for more details. Should be seq -w 1 31 as having 0 values for extra days isn't that bad
• grep $i/Nov - Gets our dates with the i variable which changes from 01 to 30 throughout the loop
• grep login - Finds lines with login; you could be a bit more exact here!
• grep POST - Finds all the requests posted.
• wc -l - Count lines in output.

I like redirecting to csv file which I can open in a spreadsheet program to produce pretty charts. An example from Office is shown below.


You can also do things like showing the top 5 login days with: sudo seq -w 1 30 | awk '{ printf $1","; system("grep " $1 "/Nov/2008 /va
r/log/apache2/ssl_access_log | grep login | grep POST | wc -l | tr \"\\n\" ,"); print "" }' | sort -t, -rnk2 | head -n5. Here sort -t, -rnk2 tells sort to use ',' as the field delimiter, to reverse the sort, to use a numeric key and to use the second field as the sort key. The head -n5 chops off the first 5 lines of input and displays it.

This is a slightly edited version which produces some command line bar charts:
sudo seq -w 1 30 | awk '{ printf $1","; system("grep " $1 "/Nov/2008 /var/log/apache2/ssl_access_log | grep login | grep POST | wc -l | tr \"\\n\" ,"); print "" }' | awk -F, '{ printf $1"\t"$2"\t"; for(i=0;i<$2;i+=100) printf "#"; print""}'

This produces something like the following.
01 1204 ############
02 1084 ##########
03 2614 ##########################
04 3068 ##############################
05 3254 ################################
06 3348 #################################
07 2806 ############################
08 1692 ################
09 1588 ###############
10 3364 #################################
11 3240 ################################
12 3106 ###############################
13 3068 ##############################
14 3178 ###############################
15 2404 ########################
16 2414 ########################
17 3496 ##################################
18 3482 ##################################
19 3336 #################################
20 3164 ###############################
21 3058 ##############################
22 2152 #####################
23 1760 #################
24 3056 ##############################
25 2908 #############################
26 2906 #############################
27 2508 #########################
28 2342 #######################
29 1650 ################
30 1448 ##############


The code to produce the graph is awk -F, '{ printf $1"\t"$2"\t"; for(i=100;i<$2;i+=100) printf "#"; print""}'. Here awk is told to recognise ',' as the field separator. In the awk program the first two fields are printed out with tab separators. We use a loop to print out each '#' you may want to adjust the magic number 100 so that the chart prints nicely on your screen.

I have left out a few things:

1. My one liner is a bit long and can be made shorter; how?
2. Can the bar charts be made with some sed voodoo?
3. Can you reorientate the bar chart?
4. Can you automatically calculate the best values for i?
5. Imagine you have 5 hosted sites (each site's URL starts /a/, /b/, /c/, /d/, /e/) rewrite the one liner to add an extra country field to show logins for each country.
6. Add your own unique one liner to this blog post.