I think this is going to be a server resources issue — how many sites are you trying to host and are there enough resources for them. Have you tried looking at the logs and talking with your hosting support?
There are definitely specific plugins that are more server intense when they update. If it’s in house that you are doing updates, you might be able to use something like Inifinte WP (a dashboard for managing updates on multiple sites), which has a setting to limit the number of connections to any one server at a time.
Thread Starter
jbr11
(@jbr11)
Bet, thanks for replying.
Several sites. Enough resources I think. But when a problematic plugin is installed, a very large increase in php processes for a sustained time.
If it’s a specific plugin, you could post to that plugin’s support tab… maybe they have some insights?
Thread Starter
jbr11
(@jbr11)
When you say “There are definitely specific plugins that are more server intense when they update.”
Intense in what respect? What are they doing during the update that makes them start up php-fpm processes?
Mostly, they are bigger plugins in terms of files and file size, but some plugins need to make db changes too. But a poorly written plugin can cause this too.
You haven’t mentioned which plugin(s) this is. Any reason for that?
If you want, we can dig in with specifics about names of plugins, your hosting (& which hosting package). But this is about all I’ll be able to offer without more specifics….
Thread Starter
jbr11
(@jbr11)
Most recently it seems at least one of these plugins may have been a problem …
Gravity Form, Ultimate Social Media and Google Analytics
Previously one or more of these might have been a problem …
Advanced Custom Fields, All in one SEO, Autoptimize, Google analytics, Gravity Forms, SSH SFTP Updater, WP Smushit
Hmmm… I wouldn’t normally expect any of those to cause much of this kind of issue.
Where is the site hosted, and on what level plan? Is this one site, or several sites on the server? Is it a high traffic site (say, more than 1500 visitors a day)? What version of PHP are you running?
Have you looked at the logs to see if maybe updates are trying to be run at the same time as backups or a WordFence scan?
Are updates trying to run at the same time other server intensive things are happening? (for instance: lots of traffic & activity on an ecommerce or LMS site).
Thread Starter
jbr11
(@jbr11)
Hello. Thanks to everyone that gave assistance. Just going an update in case anyone is interested.
We did increase the various “timeouts” mentioned. They were probably too low.
It looks like when using PHP-FPM there is a problem of TCP sockets being leaked . The symptom seems to be fin_wait state but not equivalent number of close_wait state connections freeing up the sockets. Not sure if specific to a version of PHp_FPM.
It may not be a problem if Unix Domain Sockets are used instead of TCP. Not sure.
Here is some info on the problem …
https://blog.cloudflare.com/this-is-strictly-a-violation-of-the-tcp-specification/
