Modules - PLESK

[BRJP]

Dear All,

Does anyone have:

1. Experience with the PLESK v2 Module?
2. Experience with modules that 'time out' or cause the user experience to show a web page saying connecting dropped or NO DATA RECEIVED?

In short, my PLESK v2 module sets up accounts in PLESK but then hangs - its almost as if the plugin fails to stay connected with the web page (either in the client portal of when performed in the Admin panel or when executed in Tasks) and severs connection. Click here to see a video of what happens: https://db.tt/G6MR6BQo

If you do this in the Admin Panel the same sort of things happens - the accounts are set up but it does not end cleanly.

Anyone use PLESK v2?  Any ideas?  It could be the PLESK plugin itself but this has been working before and just wondered if it could be environmental?

Brad

[BRJP]

Here is an example when you execute the provision through tasks.  The spinning disk goes round for ever and you never get a clean exit.

https://www.dropbox.com/s/43qqnx34r152yi0/Screenshot%202016-03-26%2015.27.49.png?dl=0

[hbillclient]

It could be a firewall issue or a timeout issue. It could also be that the libxml/simplexml extension is not available or not uptodate for parsing the Plesk output.

[BRJP]

Apologies, rereading my post, I don't think I was vary clear. 

The plugin works if provisioned or terminated via the CRON.  This seems to work well.  However, any of the below more 'immediate' methods do not work.  For example,

1. Provisioning immediately when the order is placed via the client portal.
2. Provisioning within the admin panel when, for example, adding the payment triggers the provisioning immediately.

In short, the cron method seems to work well but anything instantaneous (as above) means it hangs half way through creation.  The PLESK part is finished but it never leaves the HB part in the right state.

Do you think the libxml/simplexml is still relevant?

[hbillclient]

Did you check the error logs@Plesk?

Are you using Fast CGI? It could be related to Fast CGI timeout issue.

Try increasing the timeout value:

/etc/httpd/conf.d/fcgid.conf

FcgidIOTimeout 3600

OR

PHP settings in the Plesk control panel

Restart the webserver

You can also try switching PHP Support from Fast CGI to PHP CGI mode using Plesk control panel.

Maybe this would help:

https://kb.plesk.com/en/120785

Thanks.

[BRJP]

Thanks for all the pointers.  Much appreciated.

I think I am getting closer and it's to do with HB and PLESK being on the same box and the Apache Restart Value.  I think PLESK restarts Apache every time a change is made.  When adding multiple domains the Apache gets restarted multiple times and this the connection issues and errors.  The current work around (which is not full proof) is to set PLESK Apache to restart every 300 seconds or so. This works better as not restarting every time a change is made.

This is only a workaround as Apache restarts every 300 seconds.  If your PLESK updates are occurring when that 300 seconds finishes then you will as the same issues.  Ideally, you want Apache to restart x seconds AFTER changes have been detected.  But I cannot work out how to do this.

There are similar issues reported with BLESTA users who host BLESTA and PLESK on the same box.

[hbillclient]

Yes the issues with having Plesk on the same box has been documented on the blesta forums as well as on Plesk site itself. I did not expect you having Plesk & HB on the same box though.

"Apache to restart x seconds AFTER changes" would still cause issues if there are other requests made to Plesk at that time. The possible workaround would be to have apache restart gracefully as documented in the link below:

https://kb.plesk.com/en/112020

OR

Even better, split Plesk & HB onto seperate boxes. Maybe not ideal for all concerned.

Thanks.

[BRJP]

Yes the issues with having Plesk on the same box has been documented on the blesta forums as well as on Plesk site itself. I did not expect you having Plesk & HB on the same box though.

Yes, this is not optimal at all but a temporary solution for low usage shared websites.

The possible workaround would be to have apache restart gracefully as documented in the link below

This worked a treat and forgot about Apache graceful restarts.   I have enabled this (switched the Apache Restart in PLESK back to default) and then placed the server under huge load adding and deleting multiple accounts.  Absolutely worked a treat with no time outs or poor user experience such as web page disconnects or NO DATA Errors.

Thank you very much hbclient.

[hbillclient]

Welcome and good to know it worked out for you.