IP Fragment drops high in RHEL8/9
Issue
- Host is experiencing lot of oracle database alert logs on one node
- Errors include 'ERRCHK' and 'PT DMP_REQ'
- Peer addresses and port numbers are mentioned in the error messages
- The application uses large UDP packets that will be fragmented.
nstat
shows constantly increasingIpReasmFails
andIpReasmTimeout
/proc/net/sockstat
showsFRAG:
memory very close and often equal tonet.ipv4.ipfrag_high_thresh
Environment
- Red Hat Enterprise Linux 8
- Red Hat Enterprise Linux 9
- Oracle environments commonly cause IP fragmentation, but this can happen for any application that causes IP fragmentation
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.