Project

General

Profile

Actions

VM Creation #11492

closed

Please create new VM for hosting Natural Language Processing (NLP) web applications: nlp.d4science.org

Added by Massimiliano Assante about 7 years ago. Updated almost 7 years ago.

Status:
Closed
Priority:
Urgent
Assignee:
_InfraScience Systems Engineer
Target version:
Start date:
Mar 20, 2018
Due date:
Apr 10, 2018
% Done:

100%

Infrastructure:
Production
RAM:
3
CPU:
4
DiskSpace:
10
Smartgears:
Yes
Smartgears Version:
LATEST
Application Instructions:

nothing particular

Smartgears scopes:

ParthenosVO/ParthenosLab

Applications list and version:

nlphub 1.0 available at http://ftp.d4science.org/apps/hub.war

Applications context path:

hub

External storage data disk:
Smartgears auth connector:
data transfer enabled:
No
PostgreSQL database:
No
Number of VMs:
1

Description

The host name should be "nlp.d4science.org", a standard Apache Tomcat 7 (Not Smartgeared) would suffice. A SSL certificate is required as the applications would have to run over https.

The web applications deploy in the Apache Tomcat 7 would do a minimal work (the heavy part of the process is done in DataMiner) so just input and output mostly. Not sure how much RAM and CPU are necessary, keep it to the minimum acceptable.

Actions #1

Updated by Andrea Dell'Amico about 7 years ago

We take care of deploying war in the tomcat instance. No need for tomcat manager activated.

That should never happen in production.

Actions #2

Updated by Andrea Dell'Amico about 7 years ago

Also, please fill in reasonable values for ram, cpu and disk space.

Actions #4

Updated by Andrea Dell'Amico about 7 years ago

  • RAM changed from 1 to 3
  • CPU changed from 1 to 2
  • DiskSpace changed from 5 to 10

Do you know the applications context names yet or do I have to reverse proxy /?

Who do must have access to the VM?

Actions #5

Updated by Pasquale Pagano about 7 years ago

  • CPU changed from 2 to 4
  • Smartgears changed from No to Yes

I am a bit confused.
First, this application must be a SmartGears application.

Second, the application has to be released as a gCube application. It will be on Nexus. The difference is that we wish and need to update without waiting for the release closure.

Third, each single exploitation of the application is not heavy but we hope to have several users accessing it. I increased the amount of CPUs for this reason.

Actions #6

Updated by Massimiliano Assante about 7 years ago

Pasquale Pagano wrote:

I am a bit confused.
First, this application must be a SmartGears application.

If we want to use SmartGears is ok but then some context must be able to be called without token, i'm sure this is possible

Second, the application has to be released as a gCube application. It will be on Nexus. The difference is that we wish and need to update without waiting for the release closure.

Applicable for NLPHub not for the others derived from my prortytpe and developed by Clarin (Liner2, CSTLemma)

Third, each single exploitation of the application is not heavy but we hope to have several users accessing it. I increased the amount of CPUs for this reason.

Actions #7

Updated by Andrea Dell'Amico about 7 years ago

Pasquale Pagano wrote:

Third, each single exploitation of the application is not heavy but we hope to have several users accessing it. I increased the amount of CPUs for this reason.

what about load balancing it from the start? I guess that a new developed application will be load-balancer friendly.

Actions #8

Updated by Andrea Dell'Amico about 7 years ago

Massimiliano Assante wrote:

Applicable for NLPHub not for the others derived from my prortytpe and developed by Clarin (Liner2, CSTLemma)

They will never have a public distribution?

Actions #9

Updated by Massimiliano Assante about 7 years ago

Andrea Dell'Amico wrote:

Pasquale Pagano wrote:

Third, each single exploitation of the application is not heavy but we hope to have several users accessing it. I increased the amount of CPUs for this reason.

what about load balancing it from the start? I guess that a new developed application will be load-balancer friendly.

There is not load balancing need nor of 4 vCPUs need for me. All this web applications do is to take a text, construct a call (string concatenation) and invoke DataMiner which does the heavy process job, if any. Then get back an XML, parse it and display it. 4 CPUs is nonsense to do this.

Actions #10

Updated by Massimiliano Assante about 7 years ago

Andrea Dell'Amico wrote:

Massimiliano Assante wrote:

Applicable for NLPHub not for the others derived from my prortytpe and developed by Clarin (Liner2, CSTLemma)

They will never have a public distribution?

They may, but we can't assume this.

Actions #11

Updated by Andrea Dell'Amico about 7 years ago

As we discussed, all the applications should be installable from a public repository. I can create the VM in the meantime, but I don't see the advantage.

Actions #12

Updated by Massimiliano Assante about 7 years ago

  • Priority changed from High to Urgent

The NLPHub is going to be presented mid April and we need this VM up & running as soon as possible. For the time being it is going to be managed as dev (stable) machine so we're going to deploy the stable dev artifacts in there.

Actions #13

Updated by Andrea Dell'Amico about 7 years ago

I still need the application contexts and the scopes.
If you still want to install the applications manually, this one can only be a dev VM, with dev scopes and a name in the dev domain.

Actions #14

Updated by Pasquale Pagano about 7 years ago

  • Smartgears Version set to LATEST
  • Smartgears scopes updated (diff)
Actions #15

Updated by Massimiliano Assante about 7 years ago

  • Due date changed from Mar 30, 2018 to Apr 10, 2018
  • Application Instructions updated (diff)
  • Applications list and version updated (diff)
  • Applications context path updated (diff)

The VM is needed urgently by tomorrow at last, it is going to work on production on the scope indicate by Lino, you can download the artifact from here: http://ftp.d4science.org/apps/hub.war

It should respond to this URL https://nlp.d4science.org/hub

Applications context path is /hub

Actions #16

Updated by Andrea Dell'Amico about 7 years ago

  • Status changed from New to In Progress

hostname and IP address are going to be nlp.d4science.org 146.48.122.59

Actions #17

Updated by Andrea Dell'Amico about 7 years ago

There is no artifact:

root@ftp:/data/ftp/d4science/www# ls -la nlp
total 4
drwxr-xr-x  2 d4science d4science    6 2018-04-09 09:31 .
drwxr-xr-x 18 d4science d4science 4096 2018-04-09 09:31 ..
root@ftp:/data/ftp/d4science/www# ls -la nlp-hub/
total 4
drwxr-xr-x  2 d4science d4science    6 2018-04-09 09:36 .
drwxr-xr-x 18 d4science d4science 4096 2018-04-09 09:31 ..
Actions #18

Updated by Andrea Dell'Amico about 7 years ago

  • % Done changed from 0 to 70

@massimiliano.assante@isti.cnr.it let me know when it appears, all the other components have been installed.

Actions #19

Updated by Massimiliano Assante about 7 years ago

you're looking in the wrong place and wrong name:

it is downloadable from http://ftp.d4science.org/apps/hub.war

massi-laptop:~ massi$ wget http://ftp.d4science.org/apps/hub.war
--2018-04-09 15:19:36--  http://ftp.d4science.org/apps/hub.war
Resolving ftp.d4science.org... 146.48.122.130
Connecting to ftp.d4science.org|146.48.122.130|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 3513210 (3.3M) [application/java-archive]
Saving to: `hub.war'

100%[======================================>] 3,513,210    429K/s   in 9.8s    

2018-04-09 15:19:46 (350 KB/s) - `hub.war' saved [3513210/3513210]

massi-laptop:~ massi$ 

Actions #20

Updated by Andrea Dell'Amico about 7 years ago

The answer when I run the playbook was:

************************************************************************
fatal: [nlp.d4science.org]: FAILED! => {"changed": false, "dest": "/home/gcube/downloads/nlp.war", "msg": "Request failed", "response": "HTTP Error 404: Not Found", "state": "absent", "status_code": 404, "url": "http://ftp.d4science.org/apps/nlp.war"}

So, the app wasn't there or the file was not readable.

Actions #21

Updated by Massimiliano Assante about 7 years ago

Andrea Dell'Amico wrote:

The answer when I run the playbook was:

************************************************************************
fatal: [nlp.d4science.org]: FAILED! => {"changed": false, "dest": "/home/gcube/downloads/nlp.war", "msg": "Request failed", "response": "HTTP Error 404: Not Found", "state": "absent", "status_code": 404, "url": "http://ftp.d4science.org/apps/nlp.war"}

So, the app wasn't there or the file was not readable.

Nope, you used the name nlp.war instead of hub.war before

"url": "http://ftp.d4science.org/apps/nlp.war"}
Actions #22

Updated by Andrea Dell'Amico about 7 years ago

Ops. Yes.

Actions #23

Updated by Andrea Dell'Amico about 7 years ago

  • Status changed from In Progress to Feedback
  • % Done changed from 70 to 100

The app has been deployed.

Actions #24

Updated by Massimiliano Assante about 7 years ago

This site can’t be reached
nlp.d4science.org’s server IP address could not be found.
Go to http://www.d4science.org/
Search Google for nlp d4science org
ERR_NAME_NOT_RESOLVED
Actions #25

Updated by Massimiliano Assante about 7 years ago

but it works using the VPN (I'm at home) so I suppose it is a matter of minutes to make it working from my home as well.

Actions #26

Updated by Andrea Dell'Amico about 7 years ago

Yes, the DNS servers have problems updating zones to the outside since some weeks.

Actions #27

Updated by Massimiliano Assante about 7 years ago

There are a number of contexts that should be enabled (the one corresponding to the servlets of this webapp)

please see 3 url-pattern below

  <servlet-mapping>
    <servlet-name>NLPServlet</servlet-name>
    <url-pattern>/nlphub-servlet</url-pattern>
  </servlet-mapping>

  <servlet-mapping>
    <servlet-name>NLPUploader</servlet-name>
    <url-pattern>/nlphub-uploader-servlet</url-pattern>
  </servlet-mapping>

    <servlet-mapping>
    <servlet-name>NLPMapper</servlet-name>
    <url-pattern>/nlphub-mapper-servlet</url-pattern>
  </servlet-mapping>
</web-app>
Actions #28

Updated by Andrea Dell'Amico about 7 years ago

Done.

Actions #29

Updated by Massimiliano Assante about 7 years ago

Thanks, it works now. When is the SSL certificate (non self signed i mean) going to be available ?

Actions #30

Updated by Massimiliano Assante about 7 years ago

The Nlphub on production is available at http://nlp.d4science.org/hub , soon over HTTPS too.

Actions #31

Updated by Andrea Dell'Amico about 7 years ago

Massimiliano Assante wrote:

Thanks, it works now. When is the SSL certificate (non self signed i mean) going to be available ?

I really don't know. We are almost always over quota, so we'll have to wait several days. I cannot estimate it, it all depends by how many certificates are going to be renewed in the next days.

Actions #32

Updated by Andrea Dell'Amico almost 7 years ago

  • Status changed from Feedback to Closed

The certificate has been emitted on April 12th.

Actions

Also available in: Atom PDF

Add picture from clipboard (Maximum size: 8.91 MB)