Article Number: 000168182
Issue:
After upgrading a VS2 Metro-IP system running GeoSynchrony 5.x/6.0.x to GeoSynchrony 6.1.x end users may notice higher I/O response times.
The performance regression is greatest when there is zero latency between the clusters in the VS2 Metro IP system (0ms or near 0ms WAN Round Trip Time or RTT), offered load overhead latency is at times 50% higher.
A VS2 Metro-IP system with non-zero WAN latency (1+ms WAN RTT) may only experience as little as a sub-ms (microsecond) increase in write latency, but this still has the potential to impact latency-sensitive environments.
When the issue occurs the VPLEX firmware.log will have been logging floor/33 events for port show .
Example log excerpts:
128.221.253.68/cpu0/log:5988:W/"00601636d7c624637-2":5911:<5>2018/07/27 14:26:28.02: floor/33 TowerDomain domain has been stuck on pid 4720 (port show) for 150000 usecs, backtrace: none
128.221.253.68/cpu0/log:5988:W/"00601636d7c624637-2":5912:<5>2018/07/27 14:26:32.86: floor/33 TowerDomain domain has been stuck on pid 5164 (port show) for 150000 usecs, backtrace: none
128.221.253.67/cpu0/log:5988:W/"00601638320023139-2":8151:<5>2018/07/27 14:26:30.91: floor/33 TowerDomain domain has been stuck on pid 9936 (port show) for 180000 usecs, backtrace: none
VPLEX Series
VPLEX Series, VPLEX VS2
20 Nov 2020
2
Solution