Issue Huge load average, uninterruptible state

[omkarmore]

whats being accessed in the log?

The above logs mentioned were from access_ssl_log for products urls.
There is no access_log file present in logs folder or logs section.
I may have been once cleared/deleted the access_log file for some issue. Not sure
I am not getting why there is no access log shown over plesk panel.
The one i checked in /var/log/apache2 is blank

 

[omkarmore]

@Peter Debik @Dave W @john0001 Guys I tried doing strace over the php-fpm process occupying cpu.
Can you please checkout attached report and get any idea as to what might cause this issue.
Also I tried increasing hardware configuration still same issue arises.

 

[weltonw]

What is the previous PHP handler you are using, and what is the load/process count when that is being used?

 

[weltonw]

Also - why did you strace mysqld? are you seeing load increases there too?

 

[omkarmore]

What is the previous PHP handler you are using, and what is the load/process count when that is being used?

I was using 7.3 php-fpm served by apache prior.
Now using 7.4. Even if i try to switch to both still process/load count would be same.

Also - why did you strace mysqld? are you seeing load increases there too?

I was not sure but usually when this cpu is in full resource consumed state mysql would be using around 10-15% so added, it once consumed about 20% cpu, but I still think php-fpm is the culprit.

 

[weltonw]

So let's make sure I get this right.

You're currently using phpfpm 7.3. It works and resource usage is normal. Now you switch to 7.4, and it spikes? What is the resource usage difference between the two. X GB of ram? X amount of cpu?

 

[omkarmore]

So let's make sure I get this right.

You're currently using phpfpm 7.3. It works and resource usage is normal. Now you switch to 7.4, and it spikes? What is the resource usage difference between the two. X GB of ram? X amount of cpu?

No previously before 18.0.33/34 update it was working properly, after the update even if i switch between 7.3,7.4,8.0.3, in all of them I am facing same issue.
Currently I have one site running, with 2 gb ram and 1 core cpu, all of it is consumed first the cpu then slowly ram is consumed till the point where it will create swap.
Yesterday I tried to increase the hardware config to 4 gb ram 2 core processor to check in case it is a heavy site, but even in that there is 100% cpu usage.
Also when i turn the site to maintenance mode in plesk all the cpu and processes are deallocated/freed where cpu consumption reaches to around 25% and ram 30% and even less.

 

[weltonw]

> No previously before 18.0.33/34 update it was working properly, after the update even if i switch between 7.3,7.4,8.0.3, in all of them I am facing same issue.
Then what are you running currently? FastCGI? CGI? ModPHP?

 

[Dave W]

What is being accessed in the domains access_ssl_logs? You just provided the client types in the reply to my question.

 

[omkarmore]

> No previously before 18.0.33/34 update it was working properly, after the update even if i switch between 7.3,7.4,8.0.3, in all of them I am facing same issue.
Then what are you running currently? FastCGI? CGI? ModPHP?

in php info page it shows server api: FPM/FastCGI and I have selected run PHP as FPM app served by apache

 

[omkarmore]

What is being accessed in the domains access_ssl_logs? You just provided the client types in the reply to my question.

I have an ecommerce site so all products or stuff related to that are accessed,I have logs similar to this:
[20/Apr/2021:11:57:59 +0530] "GET /shop/food-beverages/foods/snacks-and-breakfast/true-elements-wheat-flakes-with-honey-and-almonds-750gm/ HTTP/1.0" 200 27098 "-" "Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/89.0.4389.130 Mobile Safari/537.36 (compatible; Googlebot/2.1; +What Is Googlebot | Google Search Central | Google Developers)"
[20/Apr/2021:11:57:06 +0530] "GET /shop/?product_tag=valve,boots,sneakers,natural,handmade,utensils,health-supports HTTP/1.0" 200 8604 "-" "Mozilla/5.0 (compatible; bingbot/2.0; +Bing Webmaster Tools)
[20/Apr/2021:11:55:47 +0530] "GET /?wc-ajax=get_refreshed_fragments HTTP/1.0" 200 1409 "-" "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/534+ (KHTML, like Gecko) BingPreview/1.0b"
[20/Apr/2021:11:55:30 +0530] "GET /shop/more/stationery/notebooks-diaries/style-tadka-things-on-my-mind/ HTTP/1.0" 200 29656 "-" "Mozilla/5.0 (compatible; SemrushBot/7~bl; +SEMrush Bot | Semrush)"

Just in case I have added following stuff in additional nginx directives for gzip compression.
gzip on;
gzip_disable "MSIE [1-6]\\.(?!.*SV1)";
gzip_proxied any;
gzip_comp_level 5;
gzip_types text/plain text/css application/javascript application/x-javascript text/xml application/xml application/rss+xml text/javascript image/x-icon image/bmp image/svg+xml;
gzip_vary on;

 

[Dave W]

Are those logs from a time that you were experiencing a high load?

 

[omkarmore]

Are those logs from a time that you were experiencing a high load?

Yes these are the logs around time when cpu is high

 

[weltonw]

> in php info page it shows server api: FPM/FastCGI and I have selected run PHP as FPM app served by apache
I thought you said that FPM doesn't work and causes your CPU load to spike?

 

[omkarmore]

> in php info page it shows server api: FPM/FastCGI and I have selected run PHP as FPM app served by apache
I thought you said that FPM doesn't work and causes your CPU load to spike?

No, I mean to say the multiple php-fpm processes are created and these processes causing 100% cpu spike. PFA screenshot

 

[omkarmore]

@Dave W @john0001 @IgorG
I just saw the changelog of 18.0.33 after which I am facing the 100% cpu spike due to php fpm processes.
Can the changes below cause the issue?
Change log related to php-fpm in 18.0.33
>>Plesk now preserves the following custom settings of the [php-fpm-pool-settings] section in additional php.ini directives: chroot, clear_env, decorate_workers_output. (PPPM-12666)
>>The official configuration of WordPress permalinks now works if nginx and PHP-FPM are used in tandem. (PPPM-12817)
>>If PHP-FPM is served by nginx in Apache+nginx hosting, custom error documents for PHP scripts now work. (PPPM-12815)
>>An add-on plan with disabled PHP settings can no longer affect the PHP settings of a subscription created based on the plan. (PPPM-12864)
I can precisely say after this march 7 update, I have been facing this issue.

 

[weltonw]

@Dave W @john0001 @IgorG
I just saw the changelog of 18.0.33 after which I am facing the 100% cpu spike due to php fpm processes.
Can the changes below cause the issue?
Change log related to php-fpm in 18.0.33
>>Plesk now preserves the following custom settings of the [php-fpm-pool-settings] section in additional php.ini directives: chroot, clear_env, decorate_workers_output. (PPPM-12666)
>>The official configuration of WordPress permalinks now works if nginx and PHP-FPM are used in tandem. (PPPM-12817)
>>If PHP-FPM is served by nginx in Apache+nginx hosting, custom error documents for PHP scripts now work. (PPPM-12815)
>>An add-on plan with disabled PHP settings can no longer affect the PHP settings of a subscription created based on the plan. (PPPM-12864)
I can precisely say after this march 7 update, I have been facing this issue.

Good catch - that's exciting- I totally missed that. Looks like the PMM I got from Plesk was different than what they put in the changelog lol.

Anyways, two things:

A. Have you changed the default plesk php settings?

B. What a
Handler are you running right now, that doesnt give you load problems?

 

[omkarmore]

Good catch - that's exciting- I totally missed that. Looks like the PMM I got from Plesk was different than what they put in the changelog lol.

Anyways, two things:

A. Have you changed the default plesk php settings?

B. What a
Handler are you running right now, that doesnt give you load problems?

A)Not before the update as far as i remember, at max I may have changed php 7.3 to 7.4 version.
B) After the issue, I tried using all possible combinations of php handlers with php versions and even tried to make nginx as primary server but in every combination I have been facing same issue.

 

[weltonw]

> After the issue, I tried using all possible combinations of php handlers with php versions and even tried to make nginx as primary server but in every combination I have been facing same issue.

So your server is non-functional right now?

 

[weltonw]

Looks like PPPM-12666 doesn't actually fix PPPM-51773 which is related. @IgorG are you able to confirm that PPPM-51773 wasn't resolved in 18.0.33/34?

(Sorry to hijack this thread)