Plesk 12.5.30-MU23 — nginx down

[Gene Steinberg]

When I try to start in the Plesk control panel, I get this:

Error: Unable to make action: Unable to manage service by nginxmng: ('start', 'nginx'). Error: [2016-02-22 08:58:34] ERR [util_exec] proc_close() failed ['/usr/local/psa/admin/bin/nginx_control' '--start'] with exit code [255]
Can not start proxy server:

Similar error when doing command line.

Need immediate help!

Peace,
Gene

 

[Gene Steinberg]

Got it working. But would like guidance for the future.

Peace,
Gene

 

[bskrakes]

What did you do to get it working?

 

[Gene Steinberg]

The host evidently figured out something. Or they had Plesk do it, but it suddenly came up about 15 minutes after their initial efforts, and mine, failed. I don't know what they did.

Peace,
Gene

 

[bskrakes]

Fair enough. Did you open a support ticket with your Hosting Provider?

 

[Gene Steinberg]

I telephoned them. They worked on this for half an hour or so. It failed, so they said they'd file a ticket with Plesk. 15-20 minutes later it was up, but I've had no explanation as to what was fixed. I'd like to know — for the future.

Peace,
Gene

 

[bskrakes]

Cool. If you can keep asking them as I'd like to know as well as I have had a few issues myself with nginx.

 

[Gene Steinberg]

The advantages seem tremendous, but I'm always worried it's so sensitive the slightest thing will bring it down. I had another one of these failures a few weeks back, but the host got it working that time, just saying they rebuild the configuration.

Peace,
Gene

 

[trialotto]

@Gene Steinberg (and @bskrakes)

Nginx is a huge complementary component to Plesk, even though the enormous power of Nginx has not really been released with the default configuration of Nginx, as provided in Plesk.

However, Nginx itself will not fail, certainly not when using the default configuration of Nginx.

Furthermore, it has nothing to do with the latest micro-updates.

It can be the case that you share some resources, for instance in the case of shared hosting or a VPS.

It can also be the case that your "hosting provider" offers you some kind of "shared Plesk instance".

Anyway, the error notification is not that worrysome, it does not particularly mean that Nginx is not working.

In essence, Nginx is a reverse proxy in the default Plesk setup and proxies for domains, hosted on a server, with the "route" to domains identified by an IP (the IP assigned to a server), a port (7080) and a header (the requested domain).

In short, if the port OR the IP is not reachable, then Nginx is still working but the proxy request can never reach it´s target: the site/domain in question is not working.

The above implies that a network failure (in the network of the "hosting provider") is a very solid explanation or likely candidate for the root cause of the problem.

If the latter is correct, you can then expect that future failures (in the internal network of the hosting provider) will have the same endresult, so be aware of that.

Hope the above helps and explains a little bit!

Regards....

 

[Gene Steinberg]

It's a dedicated server.

Peace,
Gene

 

[trialotto]

@Gene Steinberg

In the case of a dedicated server, the internal network error can still occur, even though it is very unlikely that it would cause the Nginx problem (note: I know of some situations in which the Nginx case can occur, but those situations involve a very strange network configuration).

In my humble opinion, if the issue is not present anymore, do not worry about that.

Regards...

 

[bskrakes]

I turned nginx off because I was always receive"Gateway 502 errors." I find it hard to believe that it would be a NIC issue as it is completely random. It almost seems more like a memory issue but I have more than enough memory.

Troubleshooting the issue has not been fun.

 

[trialotto]

@bskrakes,

I noticed that some other forum members have had a (I must admit, "strange") issue that can also be the root cause of the problem: fpm handlers not working fine.

Can you check the status of nginx (i.e. run the command: service nginx status) and, in addition, follow the steps as described in

https://talk.plesk.com/threads/php-...fter-12-5-30-mu-23-update.337080/#post-797158

Just let me know......feedback is always welcome.

Regards....

 

[bskrakes]

If I've got some time this week or weekend I will turn nginx back on and monitor it. During he work week I don't like turning it on and then hearing from my clients that their website has a "502 Bad Gateway Error."

 

[trialotto]

@bskrakes

If you are turning Nginx on, make sure to increase some values in the PHP settings (that is, if they are currently on default values):

- memory_limit: 256M (instead of default value 128M)
- max_execution_time: 120 (as a starting point)
- max_input_time: 120 (as a starting point)

The above steps are a requirement to have a clear distinction between (on the one hand) "pure" Nginx errors (they are very rare) and (on the other hand) Nginx error messages, that are essentially caused by PHP running out of resources (i.e. Nginx will not get a response from Apache and the Nginx proxy responds with an error code).

If you apply Nginx to one domain and the PHP settings are changed to the above, we can then exclude most of the "resource overusage related errors".

Also note that there potentially is an issue with FPM after MU23, so I would strongly recommend to activate Nginx, without activating FPM (for testing purposes).

Regards...

 

[bskrakes]

Hi trialotto,

Some of the sites which I host require FPM. I will have to do some more research on the clients using FPM.

I've made changes to one of the sites I host as per your suggestion so I will see what happens.

Thank you,