RHEL: TCP(v6) slab objects keep increasing due to application socket leak

Solution Unverified - Updated -

Issue

  • If this issue is happening the TCPv6 objects will look like following. Note how the 2nd (active_objs) and 3rd (num_objs) values are almost the same. If num_objs is much larger than active_objs. For example 20% or more. Then this could be a slab fragmentation issue. In which case see https://access.redhat.com/solutions/3380691
TCPv6              46901 46911   2176   15    8 : tunables    0    0    0 : slabdata  12000  12000      0
  • We are experiencing oom-killer issues on our blades now running RHEL7. Issues occur at about 7 days following the blades rebooting as part of a software upgrade.
  • Observed that the TCPv6 cache keeps on growing until it is huge and consuming most of the systems memory by checking the TCPv6 slab in /proc/slabinfo. However /proc/net/sockstat(6) show much less inuse value.

Environment

  • Red Hat Enterprise Linux
  • TCP or TCPv6

Subscriber exclusive content

A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.

Current Customers and Partners

Log in for full access

Log In

New to Red Hat?

Learn more about Red Hat subscriptions

Using a Red Hat product through a public cloud?

How to access this content