Please provide your feedback in this short Flings' survey.
Oct 12, 2016

Hi, I run HCIBench on stretched cluster with 8 hosts and VSAN is definitely on. Is HCI Bench supported with stretched cluster configuration? It seems not because it says VSAN not on, but datastore is accessible. Any idea?
=====================================================
2016-10-12 08:49:33 -0700: Validating VC IP and Crendetial...
2016-10-12 08:49:39 -0700: VC IP and Credential Validated
2016-10-12 08:49:39 -0700: Validating Datacenter FRA...
2016-10-12 08:49:39 -0700: Datacenter FRA Validated
2016-10-12 08:49:39 -0700: Validating Cluster FX2-streched-cluster...
2016-10-12 08:49:40 -0700: Cluster FX2-streched-cluster Validated
2016-10-12 08:49:40 -0700: Validating If Hosts in Cluster FX2-streched-cluster Could be Connected...
2016-10-12 08:50:04 -0700: All Hosts could be Connected
2016-10-12 08:50:04 -0700: Validating Network HCIBench...
2016-10-12 08:50:04 -0700: Network HCIBench Validated
2016-10-12 08:50:04 -0700: Checking If Network HCIBench is accessible from all the hosts of FX2-streched-cluster...
2016-10-12 08:50:11 -0700: Network HCIBench is accessible from all the hosts of FX2-streched-cluster
2016-10-12 08:50:11 -0700: Validating Type of Network HCIBench...
2016-10-12 08:50:11 -0700: Network HCIBench Type is Network
2016-10-12 08:50:11 -0700: Validating Datastore vsanDatastore (FX2)...
2016-10-12 08:50:12 -0700: Datastore vsanDatastore (FX2) Validated
2016-10-12 08:50:13 -0700: Checking Datastore vsanDatastore (FX2) type...
2016-10-12 08:50:13 -0700: Datastore vsanDatastore (FX2) type is vsan
2016-10-12 08:50:13 -0700: Checking If Datastore vsanDatastore (FX2) is accessible from all the hosts of FX2-streched-cluster...
2016-10-12 08:50:15 -0700: Datastore vsanDatastore (FX2) is accessible from all the hosts of FX2-streched-cluster
2016-10-12 08:50:15 -0700: Validating If VSAN Enabled in Cluster FX2-streched-cluster...
------------------------------------------------------------------------------
VSAN Is Not Enabled in Cluster FX2-streched-cluster!
------------------------------------------------------------------------------

Oct 19, 2016

Could you go to HCIBench console and run:
1. rvc 'VC_USERNAME'@VC_IP
2. input password of the vc
3. then do "cd /IP/FRA/computers/FX2-streched-cluster "
4. run "vsantest.perf.find_vsan_datastore . " and see what's the output

Thanks,
Chen

Oct 16, 2016

Will check that and let you know.

Oct 10, 2016

Any reason why the stats.html page is just blank? 0KB, the test runs with success however, I can see the metrics ramp up on all hosts.

Oct 10, 2016

Each LUN shows.
VMs = 0
IOPS = 0.00 IO/s
THROUGHPUT = 0.00 MB/s
LATENCY = 0.0000 ms
R_LATENCY = 0.0000 ms
W_LATENCY = 0.0000 ms
Testing is done, 0 out of 5 VMs finished test successfully. Please check following files for details:

However when I look at the test-status.log
Shows
Verifying if VMs are accessible
Deployment Successfully finished
I/O Test Started
I/O Test Fnished
Cleaning Up Client VMs
Testing Finished.

It looks like the test goes as intended, any idea why it is showing those null statistics ?

Oct 10, 2016

I found this on Cormacs blog --Note: The current version of the HCIbench appliance needs to reach out to the internet in order to get various fonts and css files needed to render VSAN observer graphs.

In my case I have no internet connection, can I work around this somehow?

Oct 10, 2016

it's resolved in this version.
could you check the details of the 5 files mentioned
"Testing is done, 0 out of 5 VMs finished test successfully. Please check following files for details"

Oct 11, 2016

Thanks for your reply Chen-

At the end of that log I see

"#
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGSEGV (0xb) at pc=0x00007f1171ce1112, pid=409, tid=0x00007f114efda700
#
# JRE version: Java(TM) SE Runtime Environment (8.0_102-b31) (build 1.8.0_102-b31)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (25.102-b31 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# V [libjvm.so+0x6d8112][thread 139712432891648 also had an error]
JfrStreamWriter::~JfrStreamWriter()+0x2d7a2
#
# Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# An error report file with more information is saved as:
# /root/vdbench/hs_err_pid409.log
Compiled method (nm) 2335278 666 n 0 java.io.UnixFileSystem::getBooleanAttributes0 (native)
total in heap [0x00007f115d28ec50,0x00007f115d28efb0] = 864
relocation [0x00007f115d28ed78,0x00007f115d28edc0] = 72
main code [0x00007f115d28edc0,0x00007f115d28efb0] = 496
#
# If you would like to submit a bug report, please visit:
# http://bugreport.java.com/bugreport/crash.jsp
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#
"

Oct 13, 2016

Chen, do you happen to know if a particular version of Java is required, resulting in the above error? Thanks for any input.

Oct 16, 2016

I think the Java version I'm using should be fine, could you:
1. re-run the test see if you can reproduce that
2. if you can reproduce the issue, could you send me the support files to vsanperformance@vmware.com?
a. all the logs in /opt/automation/logs
b. the result zip file by clicking "Save results" in the main page

Oct 07, 2016

I have been testing a few Deduplication array's. The new choice of being able to pre-fill with full random data is AWESOME. Thank you for that. I'm not sure exactly how it is pre-filling the data on all of the VM's, but I do know that vdbench has an option to pre-fill with a certain compress/dedup ratio data. is it possible to get this as an option on top of the fully random data. that way if we have a data set that we know has a 2:1 data reduction ratio, we can simulate that with the tests and see how effective the caching layers are?

Oct 07, 2016

HCIBench uses FIO to pre-fill data, for the random, below is the params:
[global]
ioengine=libaio
direct=1
thread
group_reporting
rw=randwrite
iodepth=8
rwmixread=0
buffer_compress_percentage=0
dedupe_percentage=0
blocksize=262144
randrepeat=0
blockalign=4096

if you want to leverage the dedup param of vdbench, you can just skip the "disk initial" config and upload your own vdbench param files with dedup setting configured. Just make sure the number of vmdks matches the number you configured on the webpage.

Or, you can generate the workload parameters from the webpage, and go to /opt/automation/vdbench-param-files to modify it w/ dedup configurations.

Oct 07, 2016

In my case I am not using DHCP and do not plan on rolling it. I set the private network to the same segment as the public. Then in the GUI of the tool, set the Network to the same and did not tick the DHCP box. Seeing Deployment started, hangs, then IP Assignment Failed.

Any workaround to this?

Oct 07, 2016

sounds like some of the client vms were not able to get ip addrs from your dhcp server, could you check that? or you can monitor the deployment process in vc with a smaller number of vms, see if they will be able to get ips from dhcp server.

Oct 10, 2016

Thanks Chen, I modified the secondary adapter of the HCI OVA, and added that into my secondary network in which DHCP is enabled, got it working.