ZeroDivisionError when running xpedite record
hello,
I was trying out the Life App after setting up Xpedite. However , I get the following error when I run xpedite record
xpedite record -p profileInfo.py [ERROR]: Detected mismatch in estimated TSC (3467770031) vs raw frequency (0) press RETURN key to, end live profile and generate report ...
scanning for samples files matching - /dev/shm/xpedite-Life-1548999708-*.data
loading counters for thread 19252 from file /dev/shm/xpedite-Life-1548999708-19252-00007ff2200a17c0.data -> 400 records | 100 txns loaded in 0.02 sec.
[ERROR]: failed to generate report
Traceback (most recent call last):
File "/home/ravi.s/mylabs/Xpedite/scripts/lib/xpedite/profiler/reportgenerator.py", line 115, in generateReport
histograms = self.generateHistograms(repo, classifier, app.runId)
File "/home/ravi.s/mylabs/Xpedite/scripts/lib/xpedite/profiler/reportgenerator.py", line 56, in generateHistograms
elapsedTimeBundles = self.analytics.buildElapsedTimeBundles(txnCollections, classifier)
File "/home/ravi.s/mylabs/Xpedite/scripts/lib/xpedite/analytics/init.py", line 48, in buildElapsedTimeBundles
lambda txnsc=txnSubCollection, txnCollection=txnCollection: TxnAggregator.groupElapsedTime(
File "/home/ravi.s/mylabs/Xpedite/scripts/lib/xpedite/util/init.py", line 38, in timeAction
retVal = delegate()
File "/home/ravi.s/mylabs/Xpedite/scripts/lib/xpedite/analytics/init.py", line 49, in
Also, when we call xpedite list, we get the following error:
xpedite list
Traceback (most recent call last):
File "/home/ravi.s/mylabs/Xpedite/scripts/bin/xpedite", line 248, in
Thanks for reporting the issue.
It appears you are running Xpedite in a cpu model, that's missing in the PMU events database. Xpedite sources PMU events from Intel's opensource site , which has events for all cpu generations. However the model "GenuineIntel-6-0C", is not found in this website.
Can you share more details about the machine, where the test runs ? Is it a physical or a virtual machine. The performance counters may not be exposed by the underlying virtual machine runtime.
The ZeroDivisionError, happens due to failure to get advertised Cpu frequency. We have plans to improve this code, to make it more accurate. As a quick workaround, I can make a quick fix to fall back to estimated TSC, in case the advertised frequency is not available.
Thanks Mani for looking taking up this issue.
The test run was done on a physical machine. Following are some other processor details:
Architecture: x86_64 Vendor ID: GenuineIntel CPU family: 6 Model: 44 Model name: Westmere E56xx/L56xx/X56xx (Nehalem-C)