"http-443-205" daemon prio=10 tid=0x096a8400 nid=0x2a04 runnable [0x3089d000..0x3089e130] java.lang.Thread.State: RUNNABLE at org.apache.tomcat.jni.Socket.sendbb(Native Method) at org.apache.coyote.http11.InternalAprOutputBuffer.flushBuffer(InternalAprOutputBuffer.java:711) at org.apache.coyote.http11.InternalAprOutputBuffer$SocketOutputBuffer.doWrite(InternalAprOutputBuffer.java:742) at org.apache.coyote.http11.filters.IdentityOutputFilter.doWrite(IdentityOutputFilter.java:127) at org.apache.coyote.http11.InternalAprOutputBuffer.doWrite(InternalAprOutputBuffer.java:552) at org.apache.coyote.Response.doWrite(Response.java:560)
We saw the same pattern as above in thread dumps from our hosts exhibiting CPU spikes, but with a notable exception. In our case whenever we captured traces, we saw at least _two_ threads in RUNNABLE state at org.apache.tomcat.jni.Socket.sendbb(). There was also some evidence that the number of threads in that state correlated with the severity of the CPU spike.
M --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org