Two days of crodaloola soaping
I worked on single soap request for two days and thats because nod32 probably had a bug or my company eset setting were inadequate. Actually I don’t even care. Virus protection should not cause two days of headache – it should avoid it.
So at start my problems revolved around executing soap request with spring ws using weblogic. A lot of message constructing and classloading issues that were resolved using google. But then, after request constructing was fixed, my request was sent and weblogic hanged!! After 10 minutes stack was printed out warning that some thread has been stuck for too long.
After replacing few soap related weblogic internal components I had an open source solution. Namely I started to use apache httpclient instead of weblogic internal connectionFactory. Now I could configure everything more easily and follow many of the google advices. I tried getting rid of Transfer-encoding: chunked, adding proxy servlets, configuring connection settings, using HTTP 1.0, etc.
During that time I also did enxtensive debugging. HttpClient successfully posted the request after that and before reading the response it tries to read response status. And thread was stuck at inputstream.read(). At this point my head could hold only char[3] -> {‘W’,'T’,'F’}
Then I noticed that my CPU is used 100% during the hang and most of it is used by ekrn system process (Eset Service). After weblogic crashed, CPU usage returned to normal. So run -> services.msc -> Eset Service -> manual seemed like the right thing to do. After restart everything worked like a charm.
If anyone has any ideas why NOD32 could not check that simple response for my message was virus free with 10 minutes please let me know.
RSS
