Intermittent Application Hangs Occur On An Exalogic Physical Or Virtual Server Due To A Race Condition That Occurs While Updating Memory Mapped Files Located Under NFS
(Doc ID 1924645.1)
Last updated on FEBRUARY 26, 2019
Applies to:Oracle Exalogic Elastic Cloud Software - Version 188.8.131.52.0 and later
Information in this document applies to any platform.
On Exalogic deployed in a Physical or Virtual configuration running Siebel or Oracle HTTP Server (Powered By Apache) as a standalone component or as part of another product such as E-Business Suite or Oracle Fusion Middleware, the following issues are encountered intermittently:
- Application processes (such as siebsvc, httpd) hang
- Compute nodes in a physical deployment hang and/or crash
- Guest vServers in a virtual deployment hang and/or crash
Upon inspecting /var/log/messages, the following logs are observed shortly before the hang or crash:
The above log messages indicate the following:
- The "flush" processes have become unresponsive
- Functions listed at the top of the Call Trace, such as "io_schedule", indicate that the process is blocked waiting to perform an I/O related activity
- Subsequent functions within the Call Trace, such as "nfs_writepages", indicate that the I/O activity is NFS related.
- Presence of other processes (such as "siebsvc" or "httpd") which have become blocked while attempting to update pages related to memory mapped files
- The process names of the other block processes may vary
- At the top of the Call Trace the function "io_schedule" shows these processes also waiting to perform an I/O related activity
- Function calls listed subsequently in the stack trace include "nfs_wb_page" (indicating that the blocked I/O activity is NFS related) and "handle_mm_fault" (indicating that the NFS activity is related to the update of a memory mapped file)
To view full details, sign in with your My Oracle Support account.
Don't have a My Oracle Support account? Click to get started!
In this Document