2114629 - PCD Classes com.sapportals.portal.pcd.gl.xfs.* as suspects on heap dump analysis

• System is crashing due to out of memory.

• When heap dump analysis is performed one or both of the following classes are suspects of causing the out of memory:

• com.sapportals.portal.pcd.gl.xfs.XfsDeltaLink

• com.sapportals.portal.pcd.gl.xfs.XfsContext

• Below are some examples on how the Problem Suspect would look like on the heap dump analysys performed by Eclipse Memory Analyzer.

n instances of "com.sapportals.portal.pcd.gl.xfs.XfsDeltaLink", loaded by "service:com.sap.tc.pcd.gl.srv" occupy X (Y%) bytes.

Keywords
com.sapportals.portal.pcd.gl.xfs.XfsDeltaLink
service:com.sap.tc.pcd.gl.srv

CSN Components
BC-PIN-PCD for "com.sapportals.portal.pcd.gl.xfs.XfsDeltaLink"

or

n instances of "com.sapportals.portal.pcd.gl.xfs.XfsContext", loaded by "service:com.sap.tc.pcd.gl.srv" occupy X (Y%) bytes.

Biggest instances:

• com.sapportals.portal.pcd.gl.xfs.XfsContext @ 0xfcd7def0 - 26.463.536 (2,24%) bytes.
• com.sapportals.portal.pcd.gl.xfs.XfsContext @ 0x100e33dd8 - 19.340.880 (1,64%) bytes.
• com.sapportals.portal.pcd.gl.xfs.XfsContext @ 0x1040bc0d0 - 15.646.488 (1,32%) bytes.
• com.sapportals.portal.pcd.gl.xfs.XfsContext @ 0xf645d320 - 13.406.808 (1,13%) bytes.

 These instances are referenced from one instance of "java.util.concurrent.ConcurrentHashMap$Segment[]", loaded by "<system class loader>"

Keywords
java.util.concurrent.ConcurrentHashMap$Segment[]
service:com.sap.tc.pcd.gl.srv
com.sapportals.portal.pcd.gl.xfs.XfsContext

CSN Components
BC-PIN-PCD for "com.sapportals.portal.pcd.gl.xfs.XfsContext"

Important! - How to identify false positives

Before applying this note it is necessary to make sure that the PCD xfs cache is really using most of the heap.

For example, your system has 4 GB of available heap and it ran out of memory generating a heap dump, that after being analyzed by Eclipse Memory Analyzer indicated that com.sapportals.portal.pcd.gl.xfs is using 363,1 MB (32%) of the 1 GB of the heap dump memory.

In a preliminary analysis this would indicate that PCD is causing the out of memory but we must check how much of the total heap available is being used by the PCD.

Note that even though the system has 4 GB available the heap dump contained only 1 GB.

The percentage used by the com.sapportals.portal.pcd.gl.xfs should be compared against the whole heap available and not to the size of the help dump.

In such case, the com.sapportals.portal.pcd.gl.xfs should not be considered the cause of the out of memory and the steps from the Resolution session of this KBA can be disconsidered.

Further investigation should be performed. In such case, the out of memory issue may have occurred when some application tried to allocate a big portion of the heap and there was not enough heap space available.

Read more...

SAP Netweaver Java

SAP NetWeaver 7.0 ; SAP NetWeaver 7.3 ; SAP NetWeaver 7.4 ; SAP NetWeaver 7.5 ; SAP enhancement package 1 for SAP NetWeaver 7.0 ; SAP enhancement package 1 for SAP NetWeaver 7.3 ; SAP enhancement package 2 for SAP NetWeaver 7.0 ; SAP enhancement package 3 for SAP NetWeaver 7.0

OOM, O.O.M., out of memory, portal, performance, settings, heap dump, hprof, HardReferenceLimit, HEAP, Xfs, cache, StartCacheOptimizer, optimizer , KBA , BC-PIN-PCD , PCD Generic Layer Services , BC-JAS-COR , Enterprise Runtime, Core J2EE Framework , SV-BO-EP , Service Delivery NW EP , Problem