KPI service stopping randomly

[maric.vladimir]

Hello,

On one of our installations KPI service is stopping, resulting in 504 Gateway Timeout message.

Restarting the service with run.py -cf restart kpi solves the issue.

We are using the latest version 2.020.24b. If it holds any significance, the installation has been upgraded with the DB splitting process, with no issues.

How to debug this issue?

Thanks in advance,

[Kal_Lam]

Hi @maric.vladimir,

Maybe you could have a look at this post (which should help you on debugging):

Kobo Install Not working

Ok, that means nginx is accessible but it cannot communicate with kpi from some reason. Look at docker logs $kobo-install>python run.py --logs Try to activate DEBUG mode. Modify config files In kobo-deployments/envfiles/kpi.txt, change KPI_DJANGO_DEBUG=False to KPI_DJANGO_DEBUG=True and in kobo-deployments/envfiles/kobocat.txt change KOBOCAT_DJANGO_DEBUG=False to KOBOCAT_DJANGO_DEBUG=True Restart containers with this command from kobo-docker directory. $kobo-docker> docker-compo…

Have a great day!

[maric.vladimir]

Thank @Kal_Lam,

here is what I found in uwsgi.log of KPI:

uwsgi.log:Internal Server Error: /service_health/
uwsgi.log:{address space usage: 351973376 bytes/335MB} {rss usage: 92557312 bytes/88MB} [pid: 120|app: 0|req: 1/3] 79.99.164.183 () {40 vars in 511 bytes} [Fri Jun 26 08:46:33 2020] GET /service_health/ => generated 330 bytes in 1914 msecs (HTTP/1.0 500) 5 headers in 175 bytes (1 switches on core 0)

So it obviously enters the unrecoverable state somehow… it is not the question of nginx not being able to access it.

Any further ideas of how to proceed?

Thanks,