[Gluster-devel] [wireshark] TODO features

[Vikhyat Umrao]

Hi Niels, 

Thanks for all your inputs and help, I have submitted a patch: 


https://code.wireshark.org/review/1833 





glusterfs: show filenames in the summary for common procedures 

With this patch we will have filename on the summary for procedures MKDIR, CREATE and LOOKUP. 




Example output: 

173 18.309307 192.168.100.3 -> 192.168.100.4 GlusterFS 224 MKDIR V330 MKDIR Call, Filename: testdir 
2606 36.767766 192.168.100.3 -> 192.168.100.4 GlusterFS 376 LOOKUP V330 LOOKUP Call, Filename: 1.txt 
2612 36.768242 192.168.100.3 -> 192.168.100.4 GlusterFS 228 CREATE V330 CREATE Call, Filename: 1.txt 

Thanks, 
Vikhyat 

----- Original Message -----

From: "Niels de Vos" <ndevos at redhat.com> 
To: "Vikhyat Umrao" <vumrao at redhat.com> 
Cc: gluster-devel at gluster.org 
Sent: Tuesday, April 29, 2014 11:16:20 PM 
Subject: Re: [Gluster-devel] [wireshark] TODO features 

On Tue, Apr 29, 2014 at 06:25:15AM -0400, Vikhyat Umrao wrote: 
> Hi, 
> 
> I am interested in TODO wireshark features for GlusterFS : 
> I can start from below given feature for one procedure: 
> => display the filename or filehandle on the summary for common procedures 

Things to get you and others prepared: 

1. go to https://forge.gluster.org/wireshark/pages/Todo 
2. login and edit the wiki page, add your name to the topic 
3. clone the wireshark repository: 
$ git clone git at forge.gluster.org:wireshark/wireshark.git 
(you have been added to the 'wireshark' group, so you should have 
push access over ssh) 
4. create a new branch for your testing 
$ git checkout -t -b wip/master/visible-filenames upstream/master 
5. make sure you have all the dependencies for compiling Wireshark 
(quite a lot are needed) 
$ ./autogen.sh 
$ ./configure --disable-wireshark 
(I tend to build only the commandline tools like 'tshark') 
$ make 
6. you should now have a ./tshark executable that you can use for 
testing 


The changes you want to make are in epan/dissectors/packet-glusterfs.c. 
For example, start with adding the name of the file/dir that is passed 
to LOOKUP. The work to dissect the data in the network packet is done in 
glusterfs_gfs3_3_op_lookup_call(). It does not really matter on how that 
function gets executed, that is more a thing for an other task (add 
support for new procedures). 

In the NFS-dissector, you can see how this is done. Check the 
implementation of the dissect_nfs3_lookup_call() function in 
epan/dissectors/packet-nfs.c. The col_append_fstr() function achieves 
what you want to do. 

Of course, you really should share your changes! Now, 'git commit' your 
change with a suitable commit message and do 

$ git push origin wip/master/visible-filenames 

Your branch should now be visible under 
https://forge.gluster.org/wireshark/wireshark. Let me know, and I'll 
give it a whirl. 

Now you've done the filename for LOOKUP, I'm sure you can think of other 
things that make sense to get displayed. 

Do ask questions and send corrections if something is missing, or not 
working as explained here. This email should probably get included in 
the projects wiki https://forge.gluster.org/wireshark/pages/Home some 
where. 

Good luck, 
Niels 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-devel/attachments/20140528/2c1b34c3/attachment.html>