Issue with template export / import from Zabbix 7.0.0 to Zabbix 7.0.0

[jhboricua]

I have two Zabbix Docker deployments of Zabbix 7.0.0, one non-prod and one for prod. The database backend is MySQL 8. The plan is to cutover hosts from our current 5.0 Zabbix deployment to this. As such, I've been cloning and modifying the stock 7.0.0 templates in the non-prod deployment to fit our monitoring needs. However, when I try to export the template from the Zabbix 7.0.0 non-prod deployment to the prod one, the import is failing with 'Unexpected server error'. I'm not seeing anything on the container logs for the zabbix frontend or backend that would shed some light on this. Any thoughts?

[Markku]

There have been a few issues in support.zabbix.com about various problems with templates lately. I'm not sure if your 7.0.0 -> 7.0.0 scenario has been in any of those, but I'd suggest testing with 7.0.2. It fixes a few major problems anyway.

Markku

[jhboricua]

I'll switch to the 7.0.2 containers and report back.

[jhboricua]

No dice after upgrading to 7.0.2. Exported from non-prod deployment after upgrade. Import to prod deployment also on 7.0.2 failed.

[Markku]

Is the error message shown within the Zabbix UI or is it the web server reporting the error?

Export+import in 7.0.2 works here.

Is the problem dependent of the complexity of the template? You may want to try to create just a simple minimal template and export+import that.

Also, check the resulting yaml (assumed) file and see it that has any really apparent problem.

How about exporting+importing just within your prod deployment for a test?

Markku

[jhboricua]

it is showing in the Zabbix UI as soon as I try to import the file. I don't see anything on the frontend or backend container logs or in the MySQL server. It appears it is any template import that fails, even the ones from the zabbix repo or just exporting/deleting/reimporting one from the same server. The only thing that has changed on this deployment is that the MySQL database was partitioned following this blog. That in itself should not be causing this though.

I'm going to do a deployment in my lab and try import/export before and after the database is partition to test if that's the culprit.

[jhboricua]

Couldn't replicate this on the lab. So went digging some more. While discussing with members on my team, it turns out the security team recently enabled the AWS WAF on application load balancers. True enough, that is what was blocking the post from succeeding, which explain why I couldn't see anything on the logs. Since this is a private load balancer, I need to apply some tags to it so that the WAF is bypassed. I validated this by opening the frontend directly on the container, bypassing the WAF, and the import does succeed.