Frequently (every 1 minute) VsanPullStatistics from <vcenter> is already running! at /root/VsanPullStatistics.pl line 76

[eschek87]

Every one minute I see the following in the /var/log/sexigraf/VsanDisksPullStatistics.log logfile:
/usr/local/share/perl/5.20.2/Log/Log4perl/Logger.pm line906 - DIE [ERROR] VsanPullStatistics from is already running! at /root/VsanPullStatistics.pl line 76

Therefore the vsan dashboards are empty when I select "last 1 hour". If I select "last 3 hours" I can see graphs.

Regards,
Stephan

[rschitz]

how many vsan clusters and hosts per cluster you got?

[eschek87]

We have 4 all flash cluster:
cluster 1: 6 hosts
cluster 2: 8 hosts
cluster 3: 14 hosts
cluster 4: 16 hosts

And one hybrid cluster with 12 hosts. All as stretched clusters.

[rschitz]

check the Pull Exec Time dashboard and give us the time the vi and vsan pull take please

[eschek87]

[rschitz]

ok it takes too long to complete. we're currently improving a lot of things but it shouldn't be that long.
What's the cpu spec of the appliance (vcpu and physical cpu underneath) ?

[eschek87]

2 vCPUs (1 socket) and the host where it runs has 36 CPUs x 2.29 GHz (2 sockets, Intel(R) Xeon(R) Gold 6140 CPU @ 2.30GHz)

[rschitz]

should be enough, let me get back to you when i'm done with the testing so you can try a new version of the puller script.

[rschitz]

could you try this version of the vsan script https://github.com/sexibytes/sexigraf/blob/dev/root/VsanPullStatistics.pl

[rschitz]

try also the new Vi script to see if the completion time is also better https://github.com/sexibytes/sexigraf/blob/dev/root/ViPullStatistics.pl

[eschek87]

Pull Exec Time now looks better for vi statistics, but for vSAN it's similar. But I still have missing graphs in the "last 1 hour" view compared to the "last 3 hours view".

[rschitz]

I found another way to increase processing speed, i'll get back to you ASAP. Thanks for your feedback!

[rschitz]

please try the latest version and let us know https://github.com/sexibytes/sexigraf/blob/8e562a798f2f8a5c58e88e06dd7871ef96fc88a2/root/VsanPullStatistics.pl

[rschitz]

Could you also send us a log extract from [INFO] Start processing vCenter to [INFO] End processing vCenter so we can have a better understanding of what is taking time to process please? send it to plot[at]sexigraf.fr if you have any security concern

[eschek87]

Hi,
I tried the new version and send you an email with the logfile. From the log I can see that it needs about 2 seconds per host and we have 56 vSAN hosts.

Regards,
Stephan

[rschitz]

2sec is way too much, i'll have a close look into the logs

[phroenips]

I'm having a similar issue on 0.99g

I only have a single two node vSAN cluster (ROBO, with remote witness appliance), and my vsan pull times are generally around 40 seconds. It exceeds 1 minute fairly regularly.

SexiGraf appliance has 2 vCPU and 8GB RAM; each physical host is dual socket Intel(R) Xeon(R) CPU E5-2678 v3 @ 2.50GHz. I'll email the logs to plot[at]sexigraf.fr for review

Thanks!
Scott

[rschitz]

Hi,
I tried the new version and send you an email with the logfile. From the log I can see that it needs about 2 seconds per host and we have 56 vSAN hosts.

Regards,
Stephan

I finally made some significant improvements with powershell parallel processing, would you be interested by some testing?

[matteowiz]

Hi,

I'm experiencing same problem after updating sexigraf from f to g version, we have 20 servers in 3 clusters and with previous version it takes an average of 30s to pull data. With new version we are over one minute and vsan charts are not displaying at all. Can I try new version you mentioned?

[rschitz]

sure, reach me at plot[at]sexigraf.fr and i'll send you a link of the latest "beta" ;)

[hekyl]

HI, same here, from f to g to Ravenholm. vsanpullstatistics are not able to finish and they run may be minutes per vsan node. We'd like to test any new beta if it'll change this behavior. We'll try to reach you asu you are suggesting above, thanks

[rschitz]

5001984