Articles in this section

[SPN9] Advance investigation : translation freeze or Systran doesn't translate

  • Updated

Objective

This article aims to describe the steps to follow when encountered a translation freeze or what to do when the Systran application is not translating anymore.

Description

When translating multiple large files or when simultaneously translating large amount of texts (via text or API) sometimes you may encounter this issue, where the translation job freezes or even after hours the file is not translated.

Steps:
1/ Check the status of the services
2/ Check the queue management for any stagnant queue
3/ Check the logs

 

 

Observations and steps

1/ Check the status of the services

From the application monitoring windows and the server terminal: See this article below to check the status of the services.

If a service is down then restart the service in question

systemctl restart [name_of_service]

If the service doesn't start or if the error is coming from a particular component, then check the logs of the component

journalctl -u [name_of_service]

For more information consult the article : [SES/SPNS]Check the status, logs of systran services and it's dependencies

If you are encountering issue on a particular language or translation resource, then check the following:

2/ Check the queue management for any stagnant queue

RabbitMQ server has it's own web interface and can be reached on http://systran:SESpassword@[SERVER]:15672

mceclip0.png

The port 15672 must be open on the SERVER firewall, check the article to know more : How to whitelist ports on firewalld

Once logged in, look for the queues

mceclip1.png

 

Look for running and idle queues: you can sort by column "total"

mceclip2.png

If there are messages in an idle queue then the queue is stagnant. You can also check for the queue ID of your translation engine via the interface to find the stagnant queue.

Link to article

[SES/SPNS] How to track or link a queue of Rabbit with it's associated TR

 

 

Next click on the stagnant queue to find the queue management items

mceclip3.png

And select 'purge' to purge messages

mceclip0.png

This should delete the messages stuck in the queue. However all the pending documents or translations job has to be restarted. So basically, users will have to redo the translation request.

If purging the stagnant queue doesn't solve the issue the consider the next step.

 

3/ Check the logs

If necessary restart the systran-ses-file-translation-consumer.service.

Show file translation consumer logs :

journalctl -u systran-ses-file-translation-consumer.service

Restart the service :

systemctl restart systran-ses-file-translation-consumer.service