DIRECT and ASYNC I/O operations on mdraid generates lots of kworker processes and causes performance issues
Issue
-
The Oracle DB is configured to use a filesystem created over lvm volume. This
LVis created on top of RAID 10mdraiddevice set up usingnvmedisks. As soon as the Oracle DB startsDIRECTandASYNCIO operations, there are large number ofkworkerprocesses created and system load average increases significantly. This also results in a severe IO performance issue.-
Oracle parameter
filesystemio_options=SETALLalso causes similar issue, since this option in DB would allow it to useDIRECTandASYNCIO operations on above filesystem, lvm volume created onmdraid. -
In some occurrences these
kworkerprocesses will keep hogging CPU usage for about 30 minutes even though there's no more running workload.
-
Environment
- Red Hat Enterprise Linux 7.4
kernel: 3.10.0-693.25.4.el7EXT4/XFSfilesystem created on lvm volume usingmdraidas aPV.mdraidwithRAID 10set up created usingnvmedisks- Oracle database configured to use
DIRECTandASYNCIO
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase of over 48,000 articles and solutions.
Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.
