|
adorai posted:How many desktops? Just curious. We only have 500 or so, and it would just stop responding occasionally during our busy logon or logoff times. VCSA is great on both of our server clusters, and the windows app is great on XenDesktop. Small, only about 100. Can't say we've ever run into that. Was the VCSA sized correctly? Were you booting up VDIs in advance of logins?
|
#
?
Mar 7, 2017 01:41
|
|
|
|
| # ? May 9, 2024 20:53 |
|
|
Internet Explorer posted:Small, only about 100. Can't say we've ever run into that. Was the VCSA sized correctly? Were you booting up VDIs in advance of logins?
|
|
#
?
Mar 7, 2017 02:06
|
|
|
Just had a fun support call after upgrading my VCSA to 6.5.0a 4944578. During the upgrade process the management gui on port 5480 went dark and stopped responding. I rebooted the VCSA and the update applied correctly but the appliance management interface continued to be unresponsive. After poking through logs for about an hour we determined that something in the upgrade script declared the disable IPv6 variable twice (once in the middle of the config, another at the very end) in the lighttp.conf file and the parser didn't like that which prevented the service from running. Dropped the last line, saved, and everything was happy. Just an FYI, I'm guessing I won't be the only one seeing that.
|
|
#
?
Mar 7, 2017 23:02
|
|
|
Crosspost: Probably not a good idea to use them...
|
|
#
?
Mar 8, 2017 08:43
|
|
|
Probably a dumb question, but is it possible to mount a VMFS 5.58 volume in ESXi 6.5 without altering it? I can't find any compatibility matrices for VMFS.
|
|
#
?
Mar 8, 2017 23:56
|
|
|
anthonypants posted:Probably a dumb question, but is it possible to mount a VMFS 5.58 volume in ESXi 6.5 without altering it? I can't find any compatibility matrices for VMFS. Yes.
|
|
#
?
Mar 9, 2017 00:29
|
|
|
Does vcenter 6.5 dropping support for esxi 5.0 actually mean those hypervisors are locked out, or just that vmware support will tell us to get lost? We have some old software that won't build on anything newer than 5.0 because of the changes in the number of empty sockets presented to solaris 10. vmware support's response has been "upgrade to a newer version of java runtime" which is kinda hard since we can't get our customers to upgrade their java, either. If I have to buy a second vcenter license to move the 5.0 build cluster off to its own environment, I will, but I'd rather keep everything in one pane of glass.
|
|
#
?
Mar 10, 2017 22:17
|
|
|
anthonypants posted:Probably a dumb question, but is it possible to mount a VMFS 5.58 volume in ESXi 6.5 without altering it? I can't find any compatibility matrices for VMFS.
|
|
#
?
Mar 10, 2017 22:26
|
|
|
Ugh, the VCSA 6.5 installer doesn't like when IPv6 is disabled? This is mega dumb https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2147908
|
|
#
?
Mar 10, 2017 23:23
|
|
|
anthonypants posted:Ugh, the VCSA 6.5 installer doesn't like when IPv6 is disabled? This is mega dumb https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2147908 Well, the easy solution is to make sure that you've got the hostname in DNS and then supply it when deploying the VCSA.
|
|
#
?
Mar 11, 2017 02:07
|
|
|
big money big clit posted:Well, the easy solution is to make sure that you've got the hostname in DNS and then supply it when deploying the VCSA.
|
|
#
?
Mar 11, 2017 03:37
|
|
|
Years ago, at my old job when I was fresh to IT, we had a contracted specialist come in to help us with some server stuff that was too hard for our staff. He asked me if I had any experience with KVM and I was like "I know what it is and we spent a few hours on it in a Linux class I took at community college, but I'd really love to learn more!" So I spent the next few hours helping him set up rackmount KVM on our servers. 
|
|
#
?
Mar 12, 2017 04:02
|
|
|
anthonypants posted:I don't think that's what the issue is! I've installed the VCSA without IPv6 a few times now and now and not had that issue.
|
|
#
?
Mar 12, 2017 04:28
|
|
|
Dr. Arbitrary posted:Years ago, at my old job when I was fresh to IT, we had a contracted specialist come in to help us with some server stuff that was too hard for our staff. He asked me if I had any experience with KVM and I was like "I know what it is and we spent a few hours on it in a Linux class I took at community college, but I'd really love to learn more!" Hahaha I've heard so many stores like this one 
|
|
#
?
Mar 13, 2017 20:06
|
|
|
goondolences to those of you running Hyper-V this week https://technet.microsoft.com/library/security/MS17-008 quote:This security update resolves vulnerabilities in Microsoft Windows. The most severe of the vulnerabilities could allow remote code execution if an authenticated attacker on a guest operating system runs a specially crafted application that causes the Hyper-V host operating system to execute arbitrary code.
|
|
#
?
Mar 15, 2017 16:34
|
|
|
Docjowles posted:goondolences to those of you running Hyper-V this week It looks like the vulnerability is with vSMB as Workstation is also affected (Assuming Workstation uses vSMB for the drag-and-drop functionality): https://www.vmware.com/security/advisories/VMSA-2017-0005.html
|
|
#
?
Mar 15, 2017 16:43
|
|
|
Docjowles posted:goondolences to those of you running Hyper-V this week 
|
|
#
?
Mar 15, 2017 17:46
|
|
|
So here's a great behavior the folks at VMware didn't bother to think out: the tools uninstaller will remove absolutely everything, including the pvscsi driver, during uninstall. This causes a few minor problems on servers where the boot volume is on the pvscsi controller per their recommendation for 2016. Thanks, support.
|
|
#
?
Mar 22, 2017 23:11
|
|
|
BangersInMyKnickers posted:So here's a great behavior the folks at VMware didn't bother to think out: the tools uninstaller will remove absolutely everything, including the pvscsi driver, during uninstall. This causes a few minor problems on servers where the boot volume is on the pvscsi controller per their recommendation for 2016.
|
|
#
?
Mar 23, 2017 01:33
|
|
|
Does the built-in boot repair thing fix that? I remember back in the day it would lock the boot to a specific controller/driver to optimize startup performance and you have to do an offline sysprep generalize of the volume to fix it. Haven't run in to this kind of garbage in years.
|
|
#
?
Mar 23, 2017 14:49
|
|
|
BangersInMyKnickers posted:So here's a great behavior the folks at VMware didn't bother to think out: the tools uninstaller will remove absolutely everything, including the pvscsi driver, during uninstall. This causes a few minor problems on servers where the boot volume is on the pvscsi controller per their recommendation for 2016. Lmao, good to know
|
|
#
?
Mar 23, 2017 15:02
|
|
|
I'm putting in a bugfix report for it. I'm sure they'll take care of it in a timely manner 
|
|
#
?
Mar 23, 2017 15:32
|
|
|
That reminds me of when I got an update popup for VMware Fusion, clicked the Upgrade button and it proceeded to pause every VM I had open while connected to a VCSA because hey why not. Thankfully that was just in my lab environment but ..
|
|
#
?
Mar 23, 2017 15:33
|
|
|
Yeah, thankfully this happened on my personal dev machine and I can just pull the hourly snapshot so its not a big deal. They were trying to get me to do this on the production exchange VM and I told them to gently caress off. The bug I am chasing looks like this:  Where most of my VMs are reporting 100% memory active despite it being wildly untrue. The behavior started with 6.5 pretty much immediately, before the guests even had a chance to reboot for tools updates, so I assume its a host issue with the way it calculates a page becoming stale/inactive and the active page stat just keeps marching up and up until it hits 100%. Support is thinking its an issue with the memory balloon driver, which smells like bullshit since there is zero memory pressure on the hosts and no balloon usage in the guests and the timing doesn't add up. Could be possible but I think its unlikely, ESXi/vCenter seems the more likely culprit. I'm rolling out 6.5b on everything this weekend per their recommendation, crossing my fingers that it does something. I've had memory alarms turned off since 6.5 because they became worthless.
|
|
#
?
Mar 23, 2017 15:48
|
|
|
Maybe they changed it to consumed rather than active? Consumed never drops unless ballooning kicks in. Like you said, balloon driver has nothing to do with active, it's just some algorithm that deems that "x pages have been touched recently" and will work even without tools iirc. theperminator fucked around with this message at 07:37 on Mar 24, 2017 |
|
#
?
Mar 24, 2017 07:35
|
|
|
|
|
#
?
Mar 24, 2017 14:10
|
|
|
theperminator posted:Maybe they changed it to consumed rather than active? Consumed never drops unless ballooning kicks in. If you drill down in to the advanced performance stats for a guest exhibiting the behavior the active memory and active memory write stats account for almost 100% of all memory allocated to the VM. It's broken.  My guess is at some point something breaks and any page that has ever been written to gets marked as permanently active and it persists until the VM restarts, then promptly comes back. Having that bar represent consumed memory would be absolutely idiotic considering the opportunistic caching the OS and application code is almost certainly doing. BangersInMyKnickers fucked around with this message at 14:58 on Mar 24, 2017 |
|
#
?
Mar 24, 2017 14:56
|
|
|
BangersInMyKnickers posted:My guess is at some point something breaks and any page that has ever been written to gets marked as permanently active and it persists until the VM restarts, then promptly comes back. Having that bar represent consumed memory would be absolutely idiotic considering the opportunistic caching the OS and application code is almost certainly doing. Guest memory management is irrelevant to the hypervisor, once a page has been touched by the guest the hypervisor considers it used by the guest until the VM is power cycled or ballooning kicks in because it has no idea what blocks have been freed by the guest. this is why if you look at the cluster overview you'll notice that memory usage doesn't go down. I've had a lot of issues with this on account of memory overprovisioning combined with busy java apps in the guest that do a lot of garbage collection and then subsequently use up every single byte of memory assigned to them over time, while the memory usage from Linux' point of view was never over 50% Still, Active shouldn't ever be that high and it's clearly a bug. theperminator fucked around with this message at 00:07 on Mar 25, 2017 |
|
#
?
Mar 25, 2017 00:03
|
|
|
In 6.0 that Guest Mem - % display (along with what the memory alarm triggers on) is based on the active memory perf stat. Since Mem granted = active mem because of whatever is broken I can't tell what stat its actually using in that first screenshot. Alarming on memory commit from the VM is completely useless since the guest memory manager is going to fill it with something unless the VM is hugely overprovisioned. 6.5b day is tomorrow morning, we'll see if it fixes any of this crap. Other hot-tip: if you update your VCSA through a mounted ISO you need to completely shut down the appliance to dismount it once its done. Otherwise, the VM will panic and restart. Good job, all star. That one happened during production hours.
|
|
#
?
Mar 25, 2017 19:07
|
|
|
BangersInMyKnickers posted:Does the built-in boot repair thing fix that? I remember back in the day it would lock the boot to a specific controller/driver to optimize startup performance and you have to do an offline sysprep generalize of the volume to fix it. Haven't run in to this kind of garbage in years.
|
|
#
?
Mar 26, 2017 15:41
|
|
|
Active memory misreport bug is still present with 6.5b. Can't wait to get jerked around with support for 6 months before they begrudgingly fix it.
|
|
#
?
Mar 27, 2017 21:42
|
|
|
Is it bad if CBT is enabled on VM but nothing is backing it up? I'm imagining a scenario sort of like leaving an active snapshot around for a couple months, but I'm not clear if CBT is similar in that way.
|
|
#
?
Mar 29, 2017 18:13
|
|
|
CBT is basically a small index file that tracks which blocks on the disk have changed to create a delta that needs to be backed at the next capture. Worst case, you have a couple hundred megabytes of ctk file alongside the disk that is marking all used blocks on the disk as changed. I doubt it will hurt anything.
|
|
#
?
Mar 29, 2017 18:38
|
|
|
BangersInMyKnickers posted:CBT is basically a small index file that tracks which blocks on the disk have changed to create a delta that needs to be backed at the next capture. Worst case, you have a couple hundred megabytes of ctk file alongside the disk that is marking all used blocks on the disk as changed. I doubt it will hurt anything. Ah ok - thanks, that makes me feel better.
|
|
#
?
Mar 29, 2017 20:47
|
|
|
I mean, I wouldn't leave a snapshot hanging around for a couple of months either way. Starts to cause disk issues and high COSTOP, and consolidating it can take for loving ever while your performance tanks.
|
|
#
?
Mar 29, 2017 23:09
|
|
|
CBT is not a snapshot.
|
|
#
?
Mar 30, 2017 02:20
|
|
|
Nice little VMware vuln in ESXi, Workstation and Fusion. https://www.vmware.com/security/advisories/VMSA-2017-0006.html
|
|
#
?
Mar 30, 2017 18:55
|
|
|
BangersInMyKnickers posted:CBT is not a snapshot. Did someone say it was?
|
|
#
?
Mar 30, 2017 19:10
|
|
|
Is there any supported method of seeding a new VSAN cluster via a usb device? We ship servers to remote sites directly from Dell and need some way to get hundreds of gigs of data onto the newly built cluster . In the past for non-vm servers we have shipped usb drives but getting a usb device to be usable storage under ESXi seems to be super hacky. I have seen methods of formatting a usb device as vmfs but it has been nothing but buggy since i have been working with it. We can install ESXi without issue by booting to little usb thumbdrives to kick off the hypervisor installer but onces it up and connected to vcenter we cant find a way to get our images to the datastore and the subsequent wsus and software deployment data required for our fild servers. Thoughts?
|
|
#
?
Apr 4, 2017 13:56
|
|
|
|
| # ? May 9, 2024 20:53 |
|
|
cr0y posted:Is there any supported method of seeding a new VSAN cluster via a usb device? We ship servers to remote sites directly from Dell and need some way to get hundreds of gigs of data onto the newly built cluster . In the past for non-vm servers we have shipped usb drives but getting a usb device to be usable storage under ESXi seems to be super hacky. I have seen methods of formatting a usb device as vmfs but it has been nothing but buggy since i have been working with it.
|
|
#
?
Apr 4, 2017 14:19
|
|