SU537: [Impact: Medium] StorageGRID SG6000s may experience higher than normal read latency on flash read cache enabled volumes while running SANtricity OS 11.70.4 and 11.70.5 versions

Views:
284
Last Updated:
8/22/2023, 9:55:06 PM

收藏

Summary

[Impact: Medium]

StorageGRID SG6000s may experience higher than normal read latency on flash read cache (FRC) enabled volumes while running SANtricity OS 11.70.4 and 11.70.5 versions.

Issue Description

In rare circumstances, StorageGRID SG6000 systems might experience higher than normal read latency on FRC enabled volumes while running SANtricity OS 11.70.4 and 11.70.5 versions. FRC is enabled and required for Cassandra volumes.

Symptom

StorageGRID appliances impacted by this may experience Cassandra read timeouts. This knowledge base article’s issue section outlines the commands that can be run on the controller shell to identify and quantify the latency.

How to Identify Exposure:

  1. From https://aiq.netapp.com, go to the E-Series Daily_CSB AutoSupport.
  2. Check the section CUMULATIVE-DRIVE-VOL-STATS.XML.
  3. Search for UserLabel="StorageGRID-obj-00" to find the statistics for the volume with the Cassandra database.
  4. Find the values for flashCacheReadHitBytes and readBytes.
  5. Check the values in flashCacheReadHitBytes and readBytes for a percentage of 80% or more.
  6. The calculation is (100*flashCacheReadHitBytes/readBytes).

Example:

<readBytes>957076294144</readBytes>
<flashCacheReadHitBytes>951014879744</flashCacheReadHitBytes>
100*951014879744/957076294144=99%

Workaround

Contact NetApp Support to determine if a downgrade of a SANtricity is necessary on StorageGRID appliances exhibiting higher than normal flash read cache latency.

Solution

Upgrade to StorageGRID Appliance 2023.08 or later.