WIUNIX ACADEMY
Linux I/O latency: connect iostat, PSI and service pressure
Page 3 of 3. Review the earlier steps and prerequisites before continuing.
YOUR SYSTEM. YOUR STEPS.
Where are you working?
Choose your OS and version. The prerequisites, commands and troubleshooting follow your environment.
USE YOUR OWN VALUES
Configure this tutorial
Set the example values for your environment; related explanations and copy panels update together. Never enter passwords, private keys or tokens. Values stay in this tab and are not sent to the server or share links. Check actual DNS, certificates and network policy before running commands.
Different IP addresses, domains or ports? Open the settings below, change the values and select “Apply to tutorial”, then copy the updated commands.
Set values for my environment4 editable values
LET’S BUILD
Verify recovery with comparable work, and reverse a failed mitigation
The diagnostic panels changed no storage or service settings, so they need no configuration rollback. If you separately applied an approved workload mitigation, compare the same application operation, demand level, pressure and device intervals. Falling I/O simply because all requests stopped is not recovery.
If latency or errors worsen, restore the recorded previous workload setting through the same owning application/scheduler and reconcile queued or retried work according to its recovery procedure. Never terminate a database, remount a filesystem or force a device reset from these counters alone. Keep unresolved backend or hardware errors with the storage owner.
For the next incident, use existing monitoring to retain application latency and interval device/PSI signals with matching timestamps. Establish baselines for this workload; five samples do not establish a capacity limit or a universal alert threshold.
Authorized terminal on the affected Linux host
date -u +%FT%TZ
LC_ALL=C iostat -d -x -m -y 1 5
if test -r /proc/pressure/io; then cat /proc/pressure/io; fiComparable application work meets its objective again, with measured interval changes and any remaining uncertainty recorded. The owning application health check remains required.
GO TO THE SOURCE
Official references
Each environment states how it was reviewed. A documentation review does not claim a hands-on test of every version.
- Linux PSI interface and scope
- PSI project: interpreting contention
- Linux cgroup v2 I/O statistics and limits
- Kernel block-device statistics
- iostat reference
- pidstat reference
- sysstat upstream FAQ and accounting requirements
- procps-ng vmstat reference
- findmnt reference
- lsblk reference
- Red Hat system performance monitoring
- Ubuntu 26.10 sysstat package
- Ubuntu release coverage
- Debian sysstat package
- Debian supported releases
- Arch Linux sysstat package
- Arch Linux: avoid partial upgrades
- Manjaro package management
- AlmaLinux release support
KEEP LEARNING