In Progress CPANEL-39058 - Nginix proxy issues

[bejbi]

Hi,

We are trying to test NEW "super-hiper-price-raising module" Nginix, and we found two serious issues, making this feature praclitally unusable:

1. Nginix proxy uses only http/1.1 (not http/2 !). Nginix should rise performance but this performance is lowered by lack of http/2 protocol (which is standard in apache).
What is the reason supporting only http1.1 in Nginix ?

2. Nginix proxy works very well. But, web user editing his Wordpress post - changes are NOT visible right after aditing. Becouse of cache in Nginix. User NEED to manualy Clear tha cache in cPanel. This feature is planned in cPanel version 100 .... it will be in Current Tier in about ... half a year ! Without this possibility for cPanel user cache is unusable ....

:wq

 

[cPRex]

Hey there! For issue 1, we have case CPANEL-37171 open with our developers to get http/2 added. I do know our team is working on that, but I can't say for sure what version that will be released with.

For issue 2, version 100 is already in Edge, so I would expect that to be in the Current tier soon, possibly within the next two weeks.

 

[bejbi]

We have new issues:

1. no gzip extension

2. previous mentioned: no http/2

( both above causes the customers are angry - this issues are shown in the results of Google Page Speed Insight - customers say that they have WORSE results than before ...)

3. Nginix make reload sometimes (without any reason, mybe it is "technical reload" or something). It is about 1-2 times a day. When it reloads ALL websites lose their SSL certificates ! So in over 60-70 seconds they displays en SSL error ("untrusted site", "fraud possibilities" ... depends on browser ...)

4. When new domain is added on any account - Nginix also reloads and it takes 60-70 seconds ! All websites on the server are down becouse of SSL cert error.

points 3 and 4 are on server with about 650 account (server: Dual Xeon, 40 core, 196 GB of DDR4 RAM, full SSD)

:wq

 

[cPRex]

@bejbi - is this using the ea-nginx package? If you're making changes to accounts, I would expect the service to need a reload to happen, and with that many domains it could take a full minute to read through and load the configuration file.

 

[bejbi]

@cPRex if it is to - it confirms that Nginx Proxy is not suitable for production use. On production servers we have many accounts. Apache reloads takes so short, that is imperceptible.

On new server in hosting company new accounts are created several times a day, and the same is with creating new domains. It is not allowed to take a break/downtime all websites of more than one minute during each domain operation. And additionaly this downtime shows an error with "fraud posibilities". Hosting company could not work this way ...

 

[cPRex]

Could you submit a ticket to our team so we can check this out directly on an affected system?

 

[bejbi]

@cPRex Thank you,

We have 5 separate tickets about this behaviour open from a week. Today it was escalated to support staff at Level 4 (Senior Technical Analyst / Development).

And finish is about to deinstall Nginix Proxy. Any remove or add domain by users causes reload of Nginix configuration which take over 60 seconds and problems for all domains on the server. We have many domains on a server. But on second server we have not so much domains (about 250 domains and subdomains), and reload is also noticeable by users :/

So we deinstall Nginix Proxy and we ware waiting for resolving this issue.

Official from ticket:
"The case number is EA-10221. It might be related to the number of total domains plus some (or all) of them using CloudFlare DNS."

:wq

 

[cPanelAnthony]

Hello! I have opened a cPanel case for this same issue so it is easier for us to track.

CPANEL-39058

We will update this thread when a solution is found, or if there are important updates.