incubator-olio-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Akara Sucharitakul <akara.sucharita...@yahoo.com>
Subject Re: Problems with olio+faban
Date Fri, 11 Jun 2010 16:17:32 GMT
Well, lets see:

a) When I use preloaded DB, addPerson requests fail because of duplicate issues.

You mean a DB from a previous run. Yes, requests will fail as these persons are already in
the DB.

b) When I load DB at the time of the run, Login requests 
fail.

This is the only real problem. You need to see why this is happening. Does the user exist?
If not, the DB loader might have failed somewhere. Otherwise it might be an SQL issue or a
cookie issue (unlikely, since you mentionted it works OK under case a).

c) When I use preloaded filestore, it works fine.

The filestore can be used over and over again.

d) 
When I load the filestore at the time of the run (for which I had
to 
nfs mount the folder on to machine 3), many images are missing and
I 
get the following error.

You do not and should not have to nfs mount. You can point the dataStorage to the system containing
your file store. The image generation will happen on that local system without going through
nfs.

-Akara



________________________________
From: Sankalp Gera <sankalp293gera@gmail.com>
To: olio-user@incubator.apache.org
Sent: Fri, June 11, 2010 4:29:43 AM
Subject: Re: Problems with olio+faban

Hi all,

I am facing problem with my faban olio setup. I see that most of the
users have successfully deployed the application and are able to drive
the workload. Can someone please help me out in this regard?

My setup is like this:

machine1-  192.168.10.13 -- Glassfish, faban1,filestore, geocoder
machine2-  192.168.10.11 --  Mysql, faban2
machine3-  192.168.10.16 --  Faban master

faban1 and faban2 represent the copies of  the final configured faban
master folder from machine3  on to machines1 and machines2. Plus,
FABAN_HOME is same for all three.
I start the faban server and direct my browser to 192.168.10.16:9980.
I do not run any faban related scripts on machine1 and machine2 manually.

The run.xml is attached. Is the setup and the configuration correct
for running the workload.
I am stuck with the following things:

a) When I use preloaded DB, addPerson requests fail because of duplicate issues.
b) When I load DB at the time of the run, Login requests fail.

c) When I use preloaded filestore, it works fine.
d) When I load the filestore at the time of the run (for which I had
to nfs mount the folder on to machine 3), many images are missing and
I get the following error.

Image: http://192.168.10.13:8080/webapp/access-artifacts/xxxx.jpg size 0.

I check the filestore for these images and they were missing.

I need to sort out these issues quickly. Any sort of help will highly
appreciated.

Thanks,
Sankalp

On Thu, Jun 10, 2010 at 5:24 PM, Sankalp Gera <sankalp293gera@gmail.com> wrote:
> Hi Shanti,
>
> The image and the filestore issues seem to be resolved now. Yes, they
> were due to permission issues only. However the other errors still bug
> me.
>
> 1)
> error: 'Multipart post did not work'
>
> I am attaching the relevant portion of the server log.
>
> Actually in the above case  I used a preloaded  database.  In this
> case all the AddPerson requests fail (i guess due to duplicate key
> issues), however AddEvent requests succeed.
>
> I also tried loading the database at the time of the run. In this case
> all the Login requests fail (probably except those for the newly
> created users during the run), whereas AddPerson requests succeed. The
> error I get is
>
> java.lang.RuntimeException: Found login prompt at index xxxx, Login as
> xxxx failed.
>
> 2)
>
> I have the following configuration for my setup:
>
> Machine1 -  Glassfish, faban
> Machine2 -  Mysql, faban
> Machine3-   Faban master
>
> I have just copied the faban folder from Machine3 to machines 1 and 2.
> I have also ensured that machines can ssh to each other without being
> prompted for password.  I have set the same location for FABAN_HOME.
> Does that complete the setup? I can see that the master tries to ssh
> to other machine and runs the agent script.  If so then what is the
> difference between 'Client' and 'agent'  in the driver tab in the gui.
>
>
>
>
> Thanks,
> Sankalp
>
>
>
>
>
> On Wed, Jun 9, 2010 at 10:51 AM, Shanti Subramanyam
> <shanti.subramanyam@gmail.com> wrote:
>> On Tue, Jun 8, 2010 at 8:49 AM, Sankalp Gera <sankalp293gera@gmail.com>
>> wrote:
>>>
>>> I am using the java version of Olio. I have deployed the application
>>> and it seems to be running fine. However, when I use faban to drive
>>> the workload I encounter some errors. The errors/warnings are listed
>>> below for a workload of 25 users with db and filestore set up for 40
>>> users.(just the initial setup)
>>>
>>>
>>> 1) I cannot load the filestore from the faban interface. This is
>>> usually what I get ... '/mnt/sdb1/filestore is not a directory' .
>>> However I have manually loaded the filestore and olio seems to be
>>> working with it perfectly. However an error is reported when I try to
>>> edit my profile.
>>>
>>
>> Not sure what you mean by the last sentence above - is this related to the
>> filestore load ? The only thing I can think of is some permission issue.
>> What user is faban running as and does that user have write permission to
>> the filestore ?
>> Can you please cut-n-paste the relevant portion of the log ? If it throws an
>> exception, please show the stack trace as well.
>>
>>>
>>> 2) I also get around 35-40 warnings pertaining to 'Image:
>>> http://192.168.10.13:8080/webapp/access-artifacts/xxxxx.jpg size 0' ..
>>> Its strange because I did not make any changes to my set up and
>>> earlier I was not getting such errors.
>>>
>>
>> I have seen these. I think this is related to the images not getting
>> uploaded correctly in the AddEvent/AddPerson operations. You can turn on
>> FINE or FINEST logging level in your appserver to track this down. We have
>> lots of debug log msgs from the java olio app.
>>
>>>
>>> 3) Also I usually get significant number of warnings reporting
>>> 'Multipart post did not work' when it tries to add a person.
>>>
>>
>> This is also a common error. Look at the server log.
>>
>>>
>>> 4) Also there are issues with copying of log and error files. It
>>> reports that files like server.log etc do not exist when actually they are
>>> there.
>>>
>> Are you using glassfish ? If not, this won't work. Even for glassfish, if
>> you have configured log files wilth non-default names, then things won't
>> work.
>>
>> Feel free to post your run.xml if you can't make progress.
>>
>>>
>>> Thanks,
>>> Sankalp
>>
>> Shanti
>>
>



      
Mime
View raw message