RHEL: One NIC TX queue is much busier than others, packet are dropped
Issue
- We are investigating an ICMP packet loss issue in our platform. Packets are lost in the path between a blade server connected to a switch and then to an external load balancer. We have not yet concluded which device causes the problem. Our traces show that the ICMP request is leaving the blade but doesn't appear in the load balancer.
- The only evidence we have in the blade, which is alarming, is the raise of tx_stops counter of a particular TX queue of the 10G interface (Emulex). This queue (txq6) also receives an order of magnitude more requests compared to the others.
Environment
- Red Hat Enterprise Linux 6
- be2net driver
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.
