t4nk196Hi guys, are there any good powercli scripts to determine which VM is abusing the network?
flatsidhi i have deployed vcenter for the first time. im wondering - im at SSO configuration stage. whats the meaning of it? i dont plan to use AD...
flatsiddo i just leave it as it is? with administrator and vsphere.local?
zboubyHi
zboubydo somebody played with dell M1000E and equalogic for vmware ?
neotichave some customers running equallogic
GVDflatsid: yes, just leave vsphere.local as your domain if you don't intend to use AD or external auth. you can set vsphere.local as the default log on domain for vcenter so you don't have to type it in again each time.
GVDalso note that root@localos might me root, it does NOT have any rights to manage vsphere.local users. as a result, you need to log on with administrator@vsphere.local to see & manage those user rights.
EaglemanDoes anyone know if its possible to change the length of a heartbeat packet?
EaglemanSeems to be caused when upgrading from 6.0 to 6.5 (esxi), fresh install uses correct length for heartbeat packets
sabestoI'm having issues with an USB device connected to a VM, spesifically an aeotec z-stick gen5 (zwave controller). it shows up correctly, and openzwave starts just fine, but after a few minutes it seems to stop receiving data. I am unsure how to debug this
sabestoany tips on settings i can try tweaking? it does not seem like my HW supports passthrough, all devices are marked as not capable (Supermicro A1SRi-2758F)
sabestoi have tried all combinations of USB 3.0/2.0 ports on the motherboard, USB 3.0/2.0 controllers to the VM, esxi is version 6.5
sabestoi suppose since its just a usb-serial device, i can pass it to the VM as a serial port..
genecsabesto: passthroughs aren't entirely reliable. consider using the USB network approach if you have that feature licensed
Vaevictusam i hallucinating, or is there an NTP server in esxi?
Vaevictus(or client)
genecVaevictus: the daemon shouldn't be listening to serve others... ?
genecVaevictus: you're aware that on Linux machines, it's common to see ntpd used strictly as a client, right?
Vaevictussure
Vaevictusthat's fine
Vaevictusi meant client
Vaevictusbut anyway, i can't find it :D
genecVaevictus: what how ?? what are you doing?
Vaevictusi think my hosts are configured with wrong timezone or something... i rebooted my pdc emu DC and ... -6hours for the whole domain! :D
genecVaevictus: smeels like you are in CST
VaevictusSi.
genecVaevictus: rule of thumb: VMs have their time set in UTC at boot
Vaevictusgenec: this is probably the first time i've rebooted this vm since i moved to esxi 6 (and my new blade chassis / SAN) so i'm guessing i don't have the host configured accurately.
genecVaevictus: should be reverse.
Vaevictus??
genecVaevictus: smells like your host is perfect and your guest OS is acting stupid and assuming the VM's time will be local not UTC
Vaevictuspossibly. i'm finding it hard to verify
genecVaevictus: in the vSphere (web or thick) Client, host, configuration, time. what do you see?
genecVaevictus: and I'm guessing this wasn't reboot but power off, power on, right??
Vaevictusshould have been a reboot... it was for updates
Vaevictushonestly, shortly after (or possibly before) this happened, my PACS system failed and i spent the next 11 hours working on that.
Vaevictusso i'm a bit fuzzy.
genecVaevictus: iirc, reboots normally don't reset the VM clock
Vaevictus:|
Vaevictusokay, so something's wrong
Vaevictusmy host here says it's 08:34
Vaevictus... UTC
Vaevictusalso, ntpq -p doesn't work, so i'm guessing not using ntp either
genecVaevictus: where do you see this time??
Vaevictusso ... vmotion everything away, maintenance mode, and fix this host ?
Vaevictusgenec: ssh console
Vaevictushttp://pastie.org/10969876
genecVaevictus: question just for you: is an ESXi host a Linux machine in any way?
Vaevictuspass!
genecVaevictus: I'm serious
Vaevictusuname says it's not
Vaevictus^_^
genecVaevictus: but you should be using the vSphere Client to stop the NTP daemon, set time CORRECTLY, configure the NTP daemon to look somewhere, then start it
epictetusVaevictus: not at all
epictetusVaevictus: "linux" refers to the linux kernel which it does not in any way/shape/form use
genecVaevictus: wtf? uname sounds like you don't understand the vmkernel and the ssh console
Vaevictusepictetus: that was my answer, though timid
epictetushowever, there are definitely some tools used that are also present in linux systems / similar to what you would see on linux
Vaevictusgenec: i don't understand your comment. :D
Vaevictusit's *nixy
genecVaevictus: ESXi is not Linux. ESXi has VMKernel, a hypervisor-only kernel with select drivers whose source is ported from Linux and busybox binaries in Linux/ELF-x86 format atop FreeBSD's Linux/ELF-x86 compatibility layer atop a POSIX-like compatibility layer with limited API implementation. ESX utilized Linux as a boot loader which then got virtualized into a special VM called the service console.
Vaevictusgenec: that would have been my supposition, by i was concerned my info was out of date
Vaevictusanyway
genecVaevictus: but you should be using the vSphere Client to stop the NTP daemon, set time CORRECTLY, configure the NTP daemon to look somewhere, then start it
Vaevictusi don't need to maintenace mode this thing?
genecVaevictus: maintenance mode would be the paranoid path, correct
Vaevictusand by vsphere client, do you mean thick client? i wasn't seeing how to get this going from the web client
Vaevictusbut now i found it
Vaevictusso, that was it. i needed "start and stop with host" in the ntp config.
genecVaevictus: either thick or web.
Vaevictusand, plausibly ironically, this was only wrong on one of my hosts... it happened to be a) the only one the upgrade didn't work on and i had to reinstall.... and b) the only one my pdc emu was on (and thus the only vm that was vulnerable to a bad cmos time...)
genecthey'll both have it but possibly under different labels
genecVaevictus: I normally prefer to use a script or host profiles to configure hosts post-install
Vaevictusgenec: i've only upgraded my hosts this one time, i'm certainly no expert, and the first two migrated on their own.
Vaevictusalso, it was at the end of a long week of preparation ... long dumb story.
Vaevictusanyway, i'm confident i know why the time was wrong on the host, and why the time was botched on the pdc emu dc, and thus network.
Vaevictusand that it's fixed.
Vaevictusgranted, i'm not excited about rebooting this vm again, but i probably should, just for good measure.
genecdidn't botch the network. just the server environment </pedant-networker>
Vaevictusgenec: sorry, "botched every workstation, and server, on this particular domain, affecting 200 some users, and possibly the catalyst to an 11 hour radiology downtime. "
Vaevictustcp was fine :D
genecVaevictus: oh I'll bet it cause general mayhem for every client
Vaevictusall i'm saying, is later that day, my CEO was in my server room.
genec"network-wide outage" is acceptable though :D
Vaevictus^_^
Vaevictusnot a situation i was happy to be in. somehow the pacs system puked it's network teaming driver... so ... i could ping it but anything more substancial wouldn't work... including rdp. (and naturally xrays, ct's and such).
genecVaevictus: OUCH
VaevictusACTION tests powering off the VM
genecVaevictus: I'd use test VMs
Vaevictusgenec: yeah. ER was diverting patients (who might need radiology urgently)...
Vaevictusgenec: too late ^_^. plus i was ready to manually observe and smack the server.
Vaevictusa bit cavalier i suppose
Vaevictusreckless, i suppose, really, considering the possibility that i could trigger the PACS outage again, feasibly.
Vaevictusanyway, i surived the power off, power on, the time was correct.
Vaevictusw32tm hasn't settled yet though
sabestogenec: USB over ethernet? its currently on a 60 day trial, so i could try. you got any tips for just enumerating the device on the host itself?
sabestoon any other linux distro it shows up as /dev/ttyACM0, if i could just pass that to the VM as a serial port, maybe that could work?
sabestoi did try "/etc/init.d/usbarbitrator stop", but i still cannot see the device, or find any trace of it in the logs ive checked
genecsabesto: USB over network requires that you have a network USB device to attach with.
genecsabesto: question just for you: is an ESXi host a Linux machine in any way?
sabestogenec: no
Oliberstop spreading those lies genec, you know it's ALL THE LINUX, ALL HAIL TORVALDS, fine, i'll go away tehn
genecOliber: :p
genecsabesto: ok. the vmkernel USB support should be limited to USB storage and USB HID. any other USB device should be passable but if your host's USB controller is unstable, that'd explain it "magically" disappearing
sabestoi need to investigate further, it does not dissapear, i just have to restart the program using it and it works again for a few minutes
genecsabesto: is there any way to check the guest OS's logs for USB disconnect/reconnect messages?
sabestogenec: about to check, a bit fiddly with the web frontend crashing when dealing with usb devices
genecsabesto: is this on a desktop-grade host?
genecsabesto: why aren't you using SSH/RDP/etc to manage the guestOS?
sabestogenec: i do use SSH, but i did disconnect the USB device from the VM to test something, and the webgui for esxi does not handle that well
sabestogenec: https://www.supermicro.nl/products/system/1U/5018/SYS-5018A-FTN4.cfm
sabestonot server grade CPU per-se
genecsabesto: well, whitebox, none the less. nice little box. I've been thinking about trying one sometime (either the 4c or 8c)...
sabestogenec: i dont see any trace of the usb device disconnecting, md5sum of lsusb before and after issue occurs is the same
genecsabesto: so this is lsusb in the Linux guestOS?
sabestoyep
genecsabesto: did you check dmesg?
sabestoyes, connected at boot, then no mention
sabestoi think i will investigate a bit more on the guestOS software side. this issue might be related to so many factors
genecsabesto: I'd also consider testing the device in a physical box with the same OS to test if it's the device flaking out
sabestoi did move it from a physical box
genecsabesto: same OS? if it was different in any way like running a different kernel revision, it could be contributing...
sabestoboth centos 7 minimal, but this open-zwave software running on top is a flaky bit of code
sabestoyeah
sabestoi havent tried, but is it possible to move the VM to my windows desktop and run in vmware workstation?
genecsabesto: sure. Workstation Pro even makes it easy to import it
sabestowould be interesting if it worked that way
genecsabesto: it does but with the caveat of needing to be a powered off VM
sabestogenec: well, it did not go that smoothly, OS cannot find the partitions. need to look at this later as i'm a bit out of time atm
genecsabesto: and what did you do? import using the Workstation import or did you half-arse copy it?
sabestoexported from esxi web frontend, opened it in vmware workstation
sabestodid try "download" directly in workstation, but that failed because of "Line 25: Unsupported hardware family 'vmx-13'."
kristian_on_linuhi ppl
kristian_on_linuokay ... by strange luck, my current virtual machine finally seems to work as it should ... a combination of updates to the guest and host, I guess
kristian_on_linushould I freeze it somehow?
sartantake an hourly vmware snapshot so you're always sure that things will work again
sartanoops was this #malicious_advice?
sartanI'm wondering if this was less vm guest/host upgrades and more 'you rebooted something that ate a lot of storage or cpu'
sartanimo, very unlikely for a particular update or patch to cause generalized performance issue