RHEL 8/9: The num_cgroups for blkio in cgroups keeps increasing

Solution Verified - Updated -

Issue

  • The num_cgroups for blkio in cgroups keeps increasing:
$ cat /proc/cgroups | grep -e subsys -e blkio | column -t
   #subsys_name  hierarchy  num_cgroups  enabled
   blkio         4          1099         1
   memory        7          28           1
  • kmalloc-* slab is consuming more memory.
0xffff8d8c80003500 kmalloc-2k                       745376 K    2048
0xffff8d8c800036c0 kmalloc-4k                       614240 K    4096
0xffff8d8c80003340 kmalloc-1k                       442416 K    1024
  • Sometimes Percpu in /proc/meminfo increase is slightly in observation.
  • Reading the /sys/fs/cgroup/blkio/blkio.throttle.io_service_bytes_recursive file can take up to several minutes and cause soft lockups (and panic/crash when enabled). The same possibly also applies to files blkio.throttle.io_serviced_recursive, blkio.bfq.io_serviced_recursive and blkio.bfq.io_service_bytes_recursive.

Environment

  • Red Hat OpenShift Container Platform (RHOCP)
    • 4.13: seen on kernel 5.14.0-284.25.1.el9_2.
    • 4.12 and before: seen on kernel kernel-4.18.0-372.52.1.el8_6 or later.
    • 4.11 and before: seen on kernel kernel-4.18.0-372.52.1.el8_6 or later.
  • Red Hat Enterprise Linux (RHEL) 8.8
    • Seen on kernel-4.18.0-477.10.1.el8_8 or later
  • Red Hat Enterprise Linux (RHEL) 8.6
    • Seen on kernel-4.18.0-372.52.1.el8_6 or later
  • cgroupV1

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