Isilon OneFS: A faulty DIMM causing Backend latency

Summary: A faulty memory module (DIMM) in a node may cause backend latency.

This article applies to This article does not apply to This article is not tied to any specific product. Not all product versions are identified in this article.
Check out other resources

Symptoms

Clients are reporting poor performance and there is high load and CPU utilization on one node due to isi_mca_dump process. 
 

# isi_for_array -s uptime
xxxxx-1:   1:37AM  up 216 days,  9:16, 1 users, load averages: 7.95, 6.69, 6.87
xxxxx-2:   1:37AM  up 216 days,  9:17, 2 users, load averages: 12.69, 7.75, 6.18
xxxxx-3:   1:37AM  up 216 days,  9:13, 0 users, load averages: 9.18, 5.96, 5.79
xxxxx-4:   1:37AM  up 128 days,  4:51, 1 users, load averages: 9.49, 6.47, 5.90
xxxxx-5:   1:37AM  up 216 days,  9:13, 0 users, load averages: 10.18, 6.25, 6.06
xxxxx-6:   1:37AM  up 216 days,  9:13, 0 users, load averages: 9.10, 6.52, 5.58
xxxxx-7:   1:37AM  up 216 days,  9:13, 0 users, load averages: 5.99, 4.45, 4.14
xxxxx-8:   1:37AM  up 90 days, 11:17, 2 users, load averages: 15.96, 26.64, 28.16 
xxxxx-9:   1:37AM  up 216 days,  9:13, 1 users, load averages: 5.99, 4.88, 5.12
xxxxx-10:  1:37AM  up 216 days,  9:13, 1 users, load averages: 12.52, 7.48, 6.28

# isi_for_array -n8 top
xxxxx-8: last pid: 75601;  load averages: 20.75, 19.19, 23.94  up 90+11:21:47    01:42:08
xxxxx-8: 135 processes: 5 running, 129 sleeping, 1 zombie
xxxxx-8:
xxxxx-8: Mem: 1310M Active, 69G Inact, 157G Wired, 96G Buf, 5078M Free
xxxxx-8: Swap:
xxxxx-8:
xxxxx-8:
xxxxx-8:   PID USERNAME       THR PRI NICE   SIZE    RES STATE   C   TIME    WCPU COMMAND
xxxxx-8: 73859 root             1 103    0   122M 10148K CPU19  19   6:00 100.00% isi_mca_dump
xxxxx-8: 74626 root             1 103    0   122M 10172K CPU26  26   3:01  98.97% isi_mca_dump

Cause

Machine Check Architecture (MCA) is an error reporting mechanism for the CPU and memory. Due to the faulty DIMM, there is a delay in the data written to or read from the affected node. This delay can cause backend latency and eventually affect the overall cluster performance. In worst-case scenarios, the latency can lead to a data unavailable (DU) situation.

Resolution

Replace the faulty DIMM.

Affected Products

Isilon

Products

Isilon, PowerScale OneFS
Article Properties
Article Number: 000069748
Article Type: Solution
Last Modified: 17 Dec 2025
Version:  6
Find answers to your questions from other Dell users
Support Services
Check if your device is covered by Support Services.