We have a web service deployed on JBoss 5.1.0 GA with JBoss Native 3.2.2.GA . (Got to know that Native WS is EOL now, will work on migration later). We used bottom-up approach and we are sending a complex java object that contains Strings, Long, Boolean and byte to web service.
We are calling web service for almost 60 to 90 times per minute. We are getting OOM GC overhead limit exceeded often (every 2 to 3 weeks), we got the heapdump (we have a heap of 1.5G) for that, on analysing the heapdump we see a lot of objects of ConcurrentHashMap for almost 1G. Most of these ConcurrentHashMaps<String, Object> are containing key as pointing to some XSD that is in-turn pointing to the web service class we have, for example the key is java.lang.String "JBossWS_webservice.my_package.my_company.com_8892700084997890448.xsd", here our web service implementation class is in com.my_company.my_package.webservice package. All the ConcurrentHashMap shows same kind of stuff.
Can anybody please advise why there are so much of ConsurrentHashMap object in heap ?
|Retrieving data ...|