Hello,

We run a multi-tenant application on Symfony v3.2.1. When we upgraded the vendors (only minor version) the application didn't respond anymore, and all our server pool reached 100% of usage (when it's <5% usually).

I tracked down, and upgraded vendor by vendor, until I upgraded from Symfony v3.2.1 to v3.2.3, as soon as the update landed the production, the same behaviour happened again.
All servers went to 100% CPU usage, and we couldn't get a single page loaded.

Even after 5 minutes after turning webserver off, all PHP threads didn't seem to be killed. Had to manually restart PHP to free all the CPU's.

Only Symfony were upgraded to reproduce the "bug" (I did it twice on production to be sure it was really a Symfony problem)

Here are some charts showing the symptoms:

• CPU load
• Application Response time and traffic during the time where Symfony 3.2.3 was on production
• Application response time 6 hours before the upgrade (running v3.2.1 for reference)

System specs; 2 servers running (database on other servers):

• PHP FPM 7.0.15-1~dotdeb+8.1
• OPcache enabled
• Nginx 1.11.9
• Xeon CPU E5-1620 v2 @ 3.70GHz
• 32Go RAM

Same server, same settings, only Symfony version changes:

Symfony v3.2.1: 6,734ms

Symfony v3.2.2: 6,574ms

Symfony v3.2.3: 19,613ms

Normal performance main request screenshot

VS Degraded performance main request screenshot

Blackfire comparaison

Here's the blackfire comparaison