Resolved 502 Bad Gateway on All domains

[jth]

502 again !

Now I will completely disable Fail2Ban for 24h.

 

[Kingsley]

502 again !

Now I will completely disable Fail2Ban for 24h.

maybe you should disable mod security and see

 

[jth]

I've said that ModSecurity was installed but never enabled ^^

 

[Kingsley]

I've said that ModSecurity was installed but never enabled ^^

I still think this is a plesk issue but they will say it's not that i should look for a system admin. i setup a plesk server on ubuntu for a friend 502 error after a day of running no modsecurity or fail2ban.

i suggest you try centos and see

 

[jth]

502 again without Fail2Ban and ModSecurity.
HEY PLESK, THIS IS A CRITICAL BUG !

 

[Kingsley]

502 again without Fail2Ban and ModSecurity.
HEY PLESK, THIS IS A CRITICAL BUG !

try in on centos 7.

also contact plesk. i think this is a bug

 

[Kingsley]

502 again !

Now I will completely disable Fail2Ban for 24h.

Contact support also

 

[IgorG]

BTW, we have updated https://kb.plesk.com/en/128431 for Ubuntu case.

 

[Kingsley]

BTW, we have updated https://kb.plesk.com/en/128431 for Ubuntu case.

ok

 

[Kingsley]

So that is how my major site was offline for 6hrs+ while i was asleep

 

[jth]

I've installed a new plesk server and the bug still present, when will you fix this critical bug ?

 

[Kingsley]

I've installed a new plesk server and the bug still present, when will you fix this critical bug ?

Still Ubuntu?

I use centos which is kinda good as the Sites was down for over 6 hours in the Last 20 days.

Try Centos 7.

This is a Bug, if they have taken me serious when i reported this in 12.5 Last year a solution would be available now

 

[jth]

Yes, it's still an ubuntu server :/

I don't like Centos-Redhat-Fedora

And of course, I can't open a ticket because my licence come from a provider..
Very disappointed to pay for a partially functional software

 

[Kingsley]

Yes, it's still an ubuntu server :/

I don't like Centos-Redhat-Fedora

And of course, I can't open a ticket because my licence come from a provider..
Very disappointed to pay for a partially functional software

You can contact support, i was able to contact them with a trial license.

I don't like centos either but so long as the site is online that is fine.

My friend's Ubuntu server is yet to shut down with 502 error.

I increased the restart interval, and then change the logrotate file but with Mpm_Event

 

[UFHH01]

Even that you both resist on stating, that this is a Plesk-related bug, you really should consider to investigate your issue a bit better:

To make this clear:

Investigating your provided informations at this thread, it is obvious, that the daily logrotation configurations still contain an apache - RELOAD - process, instead of an suggested apache - RESTART - process.

=> https://talk.plesk.com/search/search?&keywords="apache"+"logrotate"

... leads us for example to: => #4 / => #5 / => #17 / ...
​

AND the already mentioned KB - article can be found at:

=> Apache crashes on reload: seg fault or similar nasty error detected in the parent process ( KB - article 128431 )​

With all these informations, you should as well be able to notice, that "apache" itself is not at all a Plesk component and the mentioned apache - bug with the result of possible segfaults are as well not Plesk-related. Pls. see as well the mentioned links at => #17 for example!



The recommendation to solve your issue is still the same:

1. Pls. change your apache version from "mpm_event" to "mpm_prefork". ( Home > Tools & Settings > Apache Web Server )
2. Pls. set the "Apache restart interval" to a higher value, than the standard "0" seconds. ( Home > Tools & Settings > Apache Web Server )
3. Pls. change EVERY "apache2 reload" - setting to a "apache2 restart" - setting at your depending logrotation definitions under "/etc/logrotate.d/".
   

   EXAMPLE ( old ):
   
   ...
       postrotate
                   if /etc/init.d/apache2 status > /dev/null ; then \
                       /etc/init.d/apache2 reload > /dev/null; \
                   fi;
   ...
   
   
   EXAMPLE ( new ):
   
   ...
       postrotate
                   if /etc/init.d/apache2 status > /dev/null ; then \
                       /etc/init.d/apache2 restart > /dev/null; \
                   fi;
   ...

If you still experience issues/errors/problems, after you followed the recommendations, it is essential, that you provide ACTUAL informations from your log - files and don't forget your current operating system, your used Plesk version ( incl. MU ) and consider to include as well informations about your changes, that you have done to solve the issue! It will help no one ( especially not yourself! ), if you waste time with "chatting" about your issue and it will help no one as well, if you try for the umpteenth time, to make Plesk responsible for the issue.

 

[Burak Özdemir]

Even that you both resist on stating, that this is a Plesk-related bug, you really should consider to investigate your issue a bit better:

To make this clear:

Investigating your provided informations at this thread, it is obvious, that the daily logrotation configurations still contain an apache - RELOAD - process, instead of an suggested apache - RESTART - process.

=> https://talk.plesk.com/search/search?&keywords="apache"+"logrotate"

... leads us for example to: => #4 / => #5 / => #17 / ...
​

AND the already mentioned KB - article can be found at:

=> Apache crashes on reload: seg fault or similar nasty error detected in the parent process ( KB - article 128431 )​

With all these informations, you should as well be able to notice, that "apache" itself is not at all a Plesk component and the mentioned apache - bug with the result of possible segfaults are as well not Plesk-related. Pls. see as well the mentioned links at => #17 for example!



The recommendation to solve your issue is still the same:

1. Pls. change your apache version from "mpm_event" to "mpm_prefork". ( Home > Tools & Settings > Apache Web Server )
2. Pls. set the "Apache restart interval" to a higher value, than the standard "0" seconds. ( Home > Tools & Settings > Apache Web Server )
3. Pls. change EVERY "apache2 reload" - setting to a "apache2 restart" - setting at your depending logrotation definitions under "/etc/logrotate.d/".
   

   EXAMPLE ( old ):
   
   ...
       postrotate
                   if /etc/init.d/apache2 status > /dev/null ; then \
                       /etc/init.d/apache2 reload > /dev/null; \
                   fi;
   ...
   
   
   EXAMPLE ( new ):
   
   ...
       postrotate
                   if /etc/init.d/apache2 status > /dev/null ; then \
                       /etc/init.d/apache2 restart > /dev/null; \
                   fi;
   ...

If you still experience issues/errors/problems, after you followed the recommendations, it is essential, that you provide ACTUAL informations from your log - files and don't forget your current operating system, your used Plesk version ( incl. MU ) and consider to include as well informations about your changes, that you have done to solve the issue! It will help no one ( especially not yourself! ), if you waste time with "chatting" about your issue and it will help no one as well, if you try for the umpteenth time, to make Plesk responsible for the issue.

hi I am using a Plesk Onyx panel as a trial version and trying to make a decision to buy Plesk Onyx.

I was having the 502 Bad Gateway issues exactly in the mornings and restart the server and next morning again... and so on...

I have implemented the suggestions in the article you recommended. this one ---> https://kb.plesk.com/128431
a. change event to prefork
b. set apache interval to 50
c. change reload to restart

and sites begin to work without a server restart etc... now I am waiting for the next morning! ))))))))))))

sorry for my english but according this article it is not recommended to use mpm_event and "WE MUST USE mpm_prefork". Is it right? I mean --> do I missing something?

if so when can we begin to use mpm_event?

thanks

 

[UFHH01]

Hi Burak Özdemir,

if so when can we begin to use mpm_event?

Could you pls. state, what advantages do you expect when using "mpm_event" ? Are you aware, that some apache - modules ( for example "mod_php" ) are not working with "mpm_event"?

 

[Burak Özdemir]

Hi Burak Özdemir,


Could you pls. state, what advantages do you expect when using "mpm_event" ? Are you aware, that some apache - modules ( for example "mod_php" ) are not working with "mpm_event"?

Hi, in fact may main focus is not the time we will be able to use mpm_event... so to go clear, my main question is:

using mpm_prefork (with other 2 more recommended settings) will resolve the issue, right?

thanks.

 

[UFHH01]

Hi Burak Özdemir,

using mpm_prefork (with other 2 more recommended settings) will resolve the issue, right?

Correct.

 

[Burak Özdemir]

Hi Burak Özdemir,


Correct.

I had been encountering the issue for a week (I guess). Every morning I was waking up to 502 bad gateway for all domains and their subdomains. I was disabling and and then enabling the domains/subdomains for them to work again.

I am using Plesk Onyx as a trial on two identical servers (for testing purposes). Both have Plesk Firewall, mod_security (with comodo rules) and fail2ban enabled + one has mod_evasive enabled.

the suggestions in the https://talk.plesk.com/threads/502-bad-gateway-on-all-domains.339677/page-3#post-812512 resolved the issue for me. (I also added servers' own IPs to whitelist of fail2ban, additionally. Not sure if this is also a configuration for the issue to be solved.

thanks.