Troubleshooting
Problem
The FPI (Fabric Performance Impact) feature is an improved bottleneck detection mechanism which replaces the legacy bottleneckmon function. FPI helps SAN administrators quickly identify points of fabric latency and slow draining devices. The type of message indicates a measure of the latency severity. There are three messages associated with FPI and message ID which are shown below. IO_PERF_IMPACT (MAPS-1003) IO_FRAME_LOSS (MAPS-1001) IO_LATENCY_CLEAR (MAPS-1004) The IO_PERF_IMPACT message signifies the detection of some latency and there are two trigger points. The first trigger point occurs for latency conditions without frame time-outs is between 10ms to less than 80 ms (10 to 79 ms). The second trigger point is when the transmit credit counter (CRED_ZERO) remains at zero for a certain percentage of one of three sliding time windows. The percentage for each of the time windows are: 70% of a 1 second interval where CRED_ZERO is zero 50% of a 5 second interval of CRED_ZERO is zero 30% of a 10 second interval with CRED_ZERO is zero Examples of IO_PERF_IMPACT messages for both trigger points are shown below. 2018/02/03-01:08:03, [MAPS-1003], 7682, SLOT 6 | FID 128, WARNING, switch, F-Port 12/18, Condition=ALL_PORTS(DEV_LATENCY_IMPACT==IO_PERF_IMPACT), Current Value:[DEV_LATENCY_IMPACT,IO_FRAME_LOSS, 49 ms Frame Delay] <--- frame delay example 2018/02/03-01:42:03, [MAPS-1003], 7683, SLOT 6 | FID 128, WARNING, switch, F-Port 4/12, Condition=ALL_PORTS(DEV_LATENCY_IMPACT==IO_PERF_IMPACT), Current Value:[DEV_LATENCY_IMPACT,IO_PERF_IMPACT, 30.1% of 10 secs] <-- CRED_ZERO example The IO_FRAME_LOSS message indicates that severe latency has been detected and this message can be also be generated by two different triggers. The first trigger is where latency conditions without frame loss is 80 ms or more. The second trigger point occurs when class 3 transmit frames are dropped due to the hold time being exceeded on the port. Examples for messages of this type due to both triggers are shown below. 2018/02/03-02:05:03, [MAPS-1001], 8672, SLOT 6 | FID 128, CRITICAL, switch, F-Port 9/36, Condition=ALL_PORTS(DEV_LATENCY_IMPACT==IO_FRAME_LOSS), Current Value:[DEV_LATENCY_IMPACT,IO_FRAME_LOSS, 154 ms Frame Delay] <-- frame delay example 2018/02/03-02:20:03, [MAPS-1001], 8677, SLOT 6 | FID 128, CRITICAL, switch, F-Port 10/34, Condition=ALL_PORTS(DEV_LATENCY_IMPACT==IO_FRAME_LOSS), Current Value:[DEV_LATENCY_IMPACT,IO_FRAME_LOSS, 3 C3TX Timeouts] <-- Dropped frames example Finally, the IO_LATENCY_CLEAR message indicates the latency to generate an IO_PERF_IMPACT and/or IO_FRAME_LOSS message has improved by dropping below 9 ms and no further transmit frames are being discarded.
Symptom
MAPS-1001/1003/1004 messages recorded in the error log and/or RASLog,
Cause
Latency and slow drain ports
Diagnosing The Problem
Finding these messages in the error log and/or RASLog
Was this topic helpful?
Document Information
Modified date:
27 February 2023
UID
ssg1S1012111