Ari, No, I could not find the cause. I checked the source and there does not look like there was any changes in the polling code, so it maybe something else. I changed to 7.0.62 with exactly the same startup scripts/jdk and the problem goes away. It sometimes also happens on my dev box running jdk1.8.0_40/8.0.21 so its not just our production server.
Maybe someone on the tomcat list can help? Cheers Greg On 9 June 2015 at 13:41, Ari Luoma <arilu...@gmail.com> wrote: > Versions I am using: > Tomcat: 8.0.9 > > Java > java version "1.7.0_65" > OpenJDK Runtime Environment (rhel-2.5.1.2.el6_5-x86_64 u65-b17) > OpenJDK 64-Bit Server VM (build 24.65-b04, mixed mode) > > CentOS release 6.5 > > I was just wondering if updating to 8.0.21 would help, but it seems that > you are already using it. > Comment on changelog of 8.0.21: Numerous fixes to the APR/native > connector to improve robustness. (markt) > > > > On Tue, Jun 9, 2015 at 3:25 PM, Ari Luoma <arilu...@gmail.com> wrote: > >> Hi, >> >> >> Did you find a solution for this? >> -- >> First of all I wan't to say that I wasn't a tomcat-users subscriber and >> couldn't response directly to your question. >> -- >> >> I am also experiencing this issue almost every day. >> Thread count rises from about 47 to 57 and CPU usage goes high even >> though there is no users using our software. >> >> I was also able to narrow it to ajp-apr-8009-Poller. >> >> When lookin it from jvisualvm it all the time changed from Wait to >> Running and back to Wait state. >> ... >> 1:30:00.172: Running >> 1:30:01.179: Wait >> 1:30:02.171: Running >> 1:30:05.251: Wait >> 1:30:06.178: Running >> 1:30:16.183: Wait >> 1:30:17.183: Running >> 1:30:21.186: Wait >> ... >> >> CPU usage was high for a couple of hours and then dropped back to zero >> and also thread count dropped back to normal. >> >> I am using CentOS and Tomcat native library. >> We have multiple Tomcat instances running on the same machine and this >> always happens concurrently on all of them. >> >> Best Regards, >> Ari Luoma >> > >
