From 2264312f9a1e2b73685e2600b2f7504e645da127 Mon Sep 17 00:00:00 2001 From: freeze <1615081+Vicente-Cheng@users.noreply.github.com> Date: Tue, 4 Jun 2024 10:17:10 +0800 Subject: [PATCH] KB: Clarify the impacted version of filesystem trim (#63) Signed-off-by: Vicente Cheng Co-authored-by: Kiefer Chang Co-authored-by: Jillian --- .../the_potential_risk_with_fstrim.md | 19 ++++++++++++++----- 1 file changed, 14 insertions(+), 5 deletions(-) diff --git a/kb/2024-01-30/the_potential_risk_with_fstrim.md b/kb/2024-01-30/the_potential_risk_with_fstrim.md index e1616f5c..75b19d6d 100644 --- a/kb/2024-01-30/the_potential_risk_with_fstrim.md +++ b/kb/2024-01-30/the_potential_risk_with_fstrim.md @@ -11,16 +11,25 @@ tags: [harvester, rancher integration, longhorn, filesystem trim] hide_table_of_contents: false --- -:::note -This issue is already resolved by Longhorn v1.5.4, v1.6.1, v1.7.0 and later versions. -::: - - Filesystem trim is a common way to release unused space in a filesystem. However, this operation is known to cause IO errors when used with Longhorn volumes that are rebuilding. For more information about the errors, see the following issues: - Harvester: [Issue 4793](https://github.com/harvester/harvester/issues/4739) - Longhorn: [Issue 7103](https://github.com/longhorn/longhorn/issues/7103) +:::info important + +Filesystem trim was introduced in Longhorn v1.4.0 because of [Issue 836](https://github.com/longhorn/longhorn/issues/836). + +Longhorn volumes affected by the mentioned IO errors can disrupt operations in Harvester VMs that use those volumes. If you are using any of the following Harvester versions, see the instructions for risk mitigation in this article. + +| Affected Harvester Version | Built-In Longhorn Version | +|----------------------------|---------------------------| +| v1.2.0 | v1.4.3 | +| v1.2.1 | v1.4.3 | +| v1.3.0 | v1.6.0 | + +::: + ## Risks Associated with Filesystem Trim A consequence of the IO errors caused by filesystem trim is that VMs using affected Longhorn volumes become stuck. Imagine the VM is running critical applications, then becomes unavailable. This is significant because Harvester typically uses Longhorn volumes as VM disks. The IO errors will cause VMs to flap between running and paused states until volume rebuilding is completed.