|
DevNull posted:Yeah, the point is not to virtualize OSX Server. The point is to have tons of copies of OSX running on a beefy ESX box. I imagine mostly for VDI. A ton of places use OSX for OpenGL work, and now the VDI is getting into graphics work, OSX would be useful.
|
#
?
Oct 17, 2013 15:04
|
|
|
|
| # ? May 9, 2024 07:39 |
|
|
Internet Explorer posted:That is pretty much what virtualization was meant for, but it sounds like you may be in over your head. You definitely don't want to use Virtualbox. You want ESXi, HyperV, or XenServer. And if you're talking about buying a new server, it should not have a Core i series chip in it. Besides this, tell your client it's 2013 and there is effectively zero penalty for virtualization unless you are seriously doing it wrong. I guarantee you that an app the even mentions an Intel i-series chip in its requirements is not going to push the limits of VMware vs bare metal.
|
|
#
?
Oct 17, 2013 15:32
|
|
|
Misogynist posted:Something datacenter-grade (hell, even just form factor) that you can virtualize OS X is critical. This is not arguable. There are tons of Mac development shops that slave their continuous integration onto a stack of Mac Mini servers stuffed into a colo rack. Nobody likes doing this. Apple likes doing this, which is really the problem in this case. They don't think the market is viable enough to support hardware with an enterprise form factor, but they want to keep a foot in the pond by adding CI to Server with "Xcode Bots" to try to supplant the ubiquitous Jenkins setups (which is another kind of stupid, but whatever). Server is targeted at Minis. Maybe the new Pro will help, but I'd put more weight on an Apple-hosted CI with a yearly fee. DevNull posted:Yeah, the point is not to virtualize OSX Server. The point is to have tons of copies of OSX running on a beefy ESX box. I imagine mostly for VDI. A ton of places use OSX for OpenGL work, and now the VDI is getting into graphics work, OSX would be useful. This is sort of what they don't want to do -- they have 40% margins on hardware. OSX is only made to sell hardware. They learned that lesson the hard way when they allowed clones. OSX is a lot better than OS8, and there's probably a market for commodity OSX (even if it's virtualized), but it's a step they're shy about taking with a huge pile of money and a business model that's obviously working -- "want to use OSX? Buy a Mac".
|
|
#
?
Oct 17, 2013 16:15
|
|
|
Misogynist posted:It's kind of antithetical to the whole OSX experience, where Apple wants to control the whole stack from hardware on up the stack, unless Apple was suddenly willing to get into the PCoIP market. They make a ton of money on the hardware. They want to keep that going as long as possible. They don't have to work with PCoIP/Teradici at all though. They could just as easily start working with Citrix. Or beef up their own VNC server that is part of OSX. evol262 posted:This is sort of what they don't want to do -- they have 40% margins on hardware. OSX is only made to sell hardware. They learned that lesson the hard way when they allowed clones. OSX is a lot better than OS8, and there's probably a market for commodity OSX (even if it's virtualized), but it's a step they're shy about taking with a huge pile of money and a business model that's obviously working -- "want to use OSX? Buy a Mac". Yeah, they are a consumer device company. They don't really care about the enterprise. Enterprise is completely different type of game. Thin margins, lots of competition. Until Apple decides they want to be part of that game, enterprise customers don't have much of an option.
|
|
#
?
Oct 17, 2013 16:24
|
|
|
OSX VDI would cause the industry to EXPLODE. One day!
|
|
#
?
Oct 17, 2013 16:42
|
|
|
Gotta Say Horizon View 5.3 is looking pretty loving good. http://myvirtualcloud.net/?p=5490
|
|
#
?
Oct 17, 2013 17:29
|
|
|
Dilbert As gently caress posted:Gotta Say Horizon View 5.3 is looking pretty loving good. What features are you excited about?
|
|
#
?
Oct 17, 2013 17:45
|
|
|
three posted:What features are you excited about? MMR for windows 7(oh this might not be ready at release), Mirage support for View Desktops, VADC,and vSAN's, and VACI now fully supported.
|
|
#
?
Oct 17, 2013 17:51
|
|
|
Dilbert As gently caress posted:MMR for windows 7(oh this might not be ready at release), Mirage support for View Desktops, VADC,and vSAN's, and VACI now fully supported. I think it's a good point release. If Mirage pans out and independent analysis proves they fixed it at scale, that could be huge for VMware EUC long-term. VSAN is awesome, but it won't be a differentiator for View vs competitors eventually. It should be interesting to see how VMware EUC grows in the future. They have better leadership now.
|
|
#
?
Oct 17, 2013 18:07
|
|
|
three posted:They have better leadership now. Which leadership are you talking about?
|
|
#
?
Oct 17, 2013 18:09
|
|
|
DevNull posted:Which leadership are you talking about? Sanjay Poonen. Everything I hear about him is positive.
|
|
#
?
Oct 17, 2013 18:10
|
|
|
three posted:I think it's a good point release. If Mirage pans out and independent analysis proves they fixed it at scale, that could be huge for VMware EUC long-term. What I feel vSan support will be good at is the SMB space, where you hit the "want Virtual Desktops, but don't want to buy a SAN/NAS to do it" space. I'm also glad they opened up the 3D acceleration to more than just nVidia
|
|
#
?
Oct 17, 2013 18:11
|
|
|
Dilbert As gently caress posted:What I feel vSan support will be good at is the SMB space, where you hit the "want Virtual Desktops, but don't want to buy a SAN/NAS to do it" space. In my dream Utopian future, everything uses VSAN.
|
|
#
?
Oct 17, 2013 18:13
|
|
|
three posted:In my dream Utopian future, everything uses VSAN. Sadly we just aren't at the full Software defined datacenter, but hell if we aren't close.
|
|
#
?
Oct 17, 2013 18:14
|
|
|
three posted:Sanjay Poonen. Everything I hear about him is positive. Yeah, he seems pretty cool. I was just curious if there was a bad impression with the old leadership. Dilbert As gently caress posted:I'm also glad they opened up the 3D acceleration to more than just nVidia This was mostly a matter of other hardware companies writing drivers for ESX. Our virtualized graphics already works across hardware, as seen on the hosted products.
|
|
#
?
Oct 17, 2013 18:19
|
|
|
three posted:In my dream Utopian future, everything uses VSAN. "What is GlusterFS and/or Swift?"
|
|
#
?
Oct 17, 2013 18:21
|
|
|
DevNull posted:
That makes sense I guess, I always thought it was just Nvidia being Nvidia and wanting a limited exclusive for their hardware.
|
|
#
?
Oct 17, 2013 18:22
|
|
|
Dilbert As gently caress posted:That makes sense I guess, I always thought it was just Nvidia being Nvidia and wanting a limited exclusive for their hardware. That is why they push pass-through(vDGA) so much. I hate that acronym. I hate vSGA too. It is virtualized or pass-through, those are good names. They can't lock you in with virtualized, because the guest never sees an Nvidia driver or a card. It just sees the VMware device and driver. It might be backed by Nvidia, Intel, AMD, or software rendering. With a bit of virtualization magic, that backend could actually change while the guest is running. Nvidia gets their lock-in with pass-through. They can sell that because it gives you things like CUDA. The performance is better as well. That comes down on us to improve virtualized graphics.
|
|
#
?
Oct 17, 2013 18:31
|
|
|
DevNull posted:Yeah, he seems pretty cool. I was just curious if there was a bad impression with the old leadership. I think VMware EUC has been ran really poorly in the past as far as delivering and innovating (e.g. RTO acquisition, ThinApp dying out, PCoIP/Blast protocol conundrum, not focusing on the User Experience and EUC aspects of VDI but more of the server-side of VDI, etc), and it was treated almost like a red-headed stepchild that was being used just to help sell vSphere. Things like Mirage are a step in the right direction. There is so much potential. Things View Needs: - App - PCoIP with TS - Pick PCoIP or Blast, buy Teradici if they pick PCoIP - Improve Persona Management - Develop something similar to Citrix Director - Open up View to run on other hypervisors - Little things like Reverse Seamless Apps - Etc. etc. three fucked around with this message at 18:39 on Oct 17, 2013 |
|
#
?
Oct 17, 2013 18:34
|
|
|
Thinking about this 2nd VCAP I take and hopefully pass to dig into some more citrix. Any charolette VM goons hit my up at Corvettefish3r @ gmail.com I'll buy you a beer
|
|
#
?
Oct 18, 2013 02:42
|
|
|
Hey VMgoons, I have two ESXi builds (CPU/MOBO/32GB) up for sale over in the mart' if anyone is interested.
|
|
#
?
Oct 19, 2013 02:14
|
|
|
Have any of you guys had problems deploying the latest vMA appliance? Deployed one from OVF to three servers for a client. First they wouldn't take the default gateway assigned to them via the text based setup, then the Web UI wouldn't pull up at all. I spent an hour diagnosing it by poking around and finally figured out that I needed to go into /opt/vmware/etc/lighttpd/lighttpd.conf and comment out the "ssl.cipher-list" before the web UI would come up. The problem is that if I'm reading google right, this breaks PCI compliance so I won't be able to do this for some of our clients. I didn't see anyone mentioning this anywhere so I'm guessing I should probably submit this as a bug to VMware, huh? Very lame. Just wondering if anyone experienced this before.
|
|
#
?
Oct 19, 2013 03:36
|
|
|
Just killed our vCenter instance. Attempted to do a switchover in the Heartbeat Console and everything went to poo poo. Now the vpxd service is refusing to start. Edit: I could really use a hand here if anyone has the time. The following event is being logged in the Windows Application event log: quote:[VpxdLdap] Failed to search OU=Instances container. This may indicate a problem with LDAP permissions for the account running VirtualCenter, or that the schema is not compatible with this version of VirtualCenter. And the following errors are being logged in vpxd.log: code:Pile Of Garbage fucked around with this message at 17:50 on Oct 19, 2013 |
|
#
?
Oct 19, 2013 17:40
|
|
|
Does VCSA use the same license as a host-based vCenter install? That is to say, if you have an Essentials license are you able to choose between one or the other, or is VCSA a separate license? Their Essentials Plus kit page seems a little vague about what the situation is.
|
|
#
?
Oct 19, 2013 17:51
|
|
|
Martytoof posted:Does VCSA use the same license as a host-based vCenter install? That is to say, if you have an Essentials license are you able to choose between one or the other, or is VCSA a separate license? Their Essentials Plus kit page seems a little vague about what the situation is. I believe so. We have a normal "VCenter 5 Standard" license that works with VCSA. cheese-cube posted:Just killed our vCenter instance. Attempted to do a switchover in the Heartbeat Console and everything went to poo poo. Now the vpxd service is refusing to start. Check that ADAM is running or that you can connect to your AD instance.
|
|
#
?
Oct 19, 2013 19:08
|
|
|
PCjr sidecar posted:I believe so. We have a normal "VCenter 5 Standard" license that works with VCSA. No problems binding to AD however I had a poke around the ADAM instance and "OU=Instances" appears to be corrupted. Currently trying to pull a copy of the ADAM database from backup. Edit: finally managed to get it all back up and running. The issue was definitely a corrupted ADAM database as the following was logged in the ADAM event log: quote:VMwareVCMSDS (4336) VMwareVCMSDS: A bad page link (error -338) has been detected in a B-Tree (ObjectId: 8, PgnoRoot: 35) of database C:\ProgramData\VMware\VMware VirtualCenter\VMwareVCMSDS\adamntds.dit (35 => 944, 945). When I connected to the ADAM instance using Ldp and attempted to expand OU=Instances,DC=virtualcenter,DC=vmware,DC=int it returned a directory services error so the database was definitely toast. I restored a copy of the ADAM database from last nights backup, stopped all vCenter services, replaced the corrupted database with the restored one and restarted all services. Thanks for the help PCjr sidecar! Any forums upgrades you'd like (i.e. Plat, Archives, etc.)? Pile Of Garbage fucked around with this message at 21:42 on Oct 19, 2013 |
|
#
?
Oct 19, 2013 20:53
|
|
|
cheese-cube posted:Thanks for the help PCjr sidecar! Any forums upgrades you'd like (i.e. Plat, Archives, etc.)?
|
|
#
?
Oct 20, 2013 06:15
|
|
|
PCjr sidecar posted:No problem, glad I could help. Plat would be awesome! Cool, what's your e-mail address?
|
|
#
?
Oct 20, 2013 06:21
|
|
|
cheese-cube posted:Cool, what's your e-mail address? pcjrsidecar@gmail.com
|
|
#
?
Oct 20, 2013 13:47
|
|
|
Gift cert sent  Back on topic, does anyone have any advice regarding JVM heap size configuration for the vCenter services? Are the settings suggested in the Hardware Requirements for vCenter Server, vCenter Single Sign On, vSphere Client, and vSphere Web Client page of the vSphere 5.1 Documentation Centre a good approximation? The reason I ask is that the numpty who originally deployed our vCenter server only gave it 8GB of memory configured the JVM heap sizes along the lines of the "Medium inventory (100-400 hosts or 1000-4000 virtual machines)" recommendation in the aforementioned article. This means that physical memory usage is always >95% which is causing excessive paging and even triggering resource exhaustion alerts. I spoke to one of our infrastructure architect guys about it and he recommended the following JVM heap sizes: quote:vCenter Inventory Service: 3GB Our environment currently has 40 hosts and 420 VMs. Would the settings above be sufficient for an environment of that size while still allowing adequate headroom for growth? It's weird that the infra architect went out of his way to specify the above sizes instead of choosing nice round-GB values so I just want to see whether he put any thought into it.
|
|
#
?
Oct 20, 2013 15:44
|
|
|
I don't know how he got those numbers but a 896MB JVM for the web server, and 1GB for vCenter, will make it run like piss rear end. Something I have seen reduce the Memory usage on Windows vCenters is installing the full Java runtime on the server, I've noticed that id decently lowered the usage for 5.1, and 5.1 u1 reduced it further. What Build number of 5.1 are you running? I'd probably add more Vram prior to reseting JVM sizes. Is there a reason you can't just up the vram? Dilbert As FUCK fucked around with this message at 22:21 on Oct 20, 2013 |
|
#
?
Oct 20, 2013 22:16
|
|
|
Dilbert As gently caress posted:I don't know how he got those numbers but a 896MB JVM for the web server, and 1GB for vCenter, will make it run like piss rear end. We are running build 1064983. When you say "installing the full Java runtime on the server" do you mean the JRE or JDK? It's already got JRE 6u30 installed (Not sure what for exactly, also yeah I know JRE 6 is EOL, we are still trying to put together our testing plan with all the application teams to roll our JRE 7 enterprise wide). I did initially suggest increasing the memory from 8GB to 10GB as the hardware requirements I linked before state that if you are running vCenter, Inventory Service and SSO on the same server you need 10GB minimum. However the infrastructure architect guy dismissed that by saying that while used memory is 95% the active memory is only 55% so increasing the servers memory wouldn't resolve the issue (Which I disagree with but what can I say, he's the "Infrastructure Architect"). I might try to get around this whole thing by raising the change request for it with the JVM heap sizes as default instead of his stupid values and then let TAG decide on whether it's the best way to go.
|
|
#
?
Oct 20, 2013 23:44
|
|
|
cheese-cube posted:We are running build 1064983. When you say "installing the full Java runtime on the server" do you mean the JRE or JDK? It's already got JRE 6u30 installed (Not sure what for exactly, also yeah I know JRE 6 is EOL, we are still trying to put together our testing plan with all the application teams to roll our JRE 7 enterprise wide). JRE, seems like you are running u1 as well. The full JRE 6 can reduce some memory overhead of the JVM's as I have observed, JRE 7 has little to no effect. quote:I did initially suggest increasing the memory from 8GB to 10GB as the hardware requirements I linked before state that if you are running vCenter, Inventory Service and SSO on the same server you need 10GB minimum. However the infrastructure architect guy dismissed that by saying that while used memory is 95% the active memory is only 55% so increasing the servers memory wouldn't resolve the issue (Which I disagree with but what can I say, he's the "Infrastructure Architect"). I have no idea where he is getting that information, limiting JVM's that are asking for an amount of ram to run effectively is going to cause you a bunch of poo poo. Honestly, Limiting the JVM settings would be my last resort. On your environment I would be saying 12GB and 2 vCPU's(maybe 3) for vCenter. loving around with the JVM size is really the last thing you should do unless you are constrained on resources. Also is by chance SQL running on the same VM? If so SQL might be trying to use as much ram as possible, I have found limiting the amount of ram it uses does help(by default it caches a very large amount of system ram), this can be done in Studio Manager. Hell I have one client running 60VM's 4 hosts, stretched cluster + Veeam. He needs at least, 6Gb for vCenter, and 8GB for backups n poo poo to run. Because people seem to find it neccisary to install BE 2012 or Veeam on the same server of vCenter in all cases. quote:I might try to get around this whole thing by raising the change request for it with the JVM heap sizes as default instead of his stupid values and then let TAG decide on whether it's the best way to go. If you have the excess ram I would bump it up to 12GB, if you still hit the 90% I would then look into going into the JVM. Dilbert As FUCK fucked around with this message at 00:34 on Oct 21, 2013 |
|
#
?
Oct 21, 2013 00:31
|
|
|
So excited - the Stanly CC VMware class starts tomorrow and I'm pretty psyched about the whole deal. Can't wait to dive in and get my learning on.
|
|
#
?
Oct 21, 2013 01:30
|
|
|
Dilbert As gently caress posted:I have no idea where he is getting that information, limiting JVM's that are asking for an amount of ram to run effectively is going to cause you a bunch of poo poo. Honestly, Limiting the JVM settings would be my last resort. On your environment I would be saying 12GB and 2 vCPU's(maybe 3) for vCenter. loving around with the JVM size is really the last thing you should do unless you are constrained on resources. Also is by chance SQL running on the same VM? If so SQL might be trying to use as much ram as possible, I have found limiting the amount of ram it uses does help(by default it caches a very large amount of system ram), this can be done in Studio Manager. Nah the vCenter and VUM databases are on a separate, clustered SQL Server instance (About the only thing that appears to have been done right). I agree that increasing the memory is the best way to go. It's alot easier and safer than dicking around with multiple wrapper.conf files (Especially as we are running vCenter Heartbeat so there's the extra rigmarole of ensuring that the changes on the active server properly replicate to the passive server). Dilbert As gently caress posted:Hell I have one client running 60VM's 4 hosts, stretched cluster + Veeam. He needs at least, 6Gb for vCenter, and 8GB for backups n poo poo to run. Because people seem to find it neccisary to install BE 2012 or Veeam on the same server of vCenter in all cases. Ahahaha this practice was prevalent as hell in my last job (Mainly because I was dealing with SMB instead of enterprise customers). We were an IBM shop so the bog standard build always ended up as a DS3550 SAN, 2 x SAN24B-4 FC switches, 3 x x3550 servers as ESXi hosts and an x3650 stuffed full of disks hooked up to a lovely LTO4 tape drive via SAS and running vCenter, BE and sometimes Veeam. Ah, simpler times.
|
|
#
?
Oct 21, 2013 02:07
|
|
|
gleh
Dilbert As FUCK fucked around with this message at 04:57 on Oct 21, 2013 |
|
#
?
Oct 21, 2013 02:42
|
|
|
Daylen Drazzi posted:So excited - the Stanly CC VMware class starts tomorrow and I'm pretty psyched about the whole deal. Can't wait to dive in and get my learning on. Also giddy with excitement, now that I'm able to actually take the class since I "moved" to Wisconsin. 
|
|
#
?
Oct 21, 2013 03:18
|
|
|
FISHMANPET posted:Also giddy with excitement, now that I'm able to actually take the class since I "moved" to Wisconsin. I was out of the country on vacation when and missed my registration window  Back on the waiting list I go.
|
|
#
?
Oct 21, 2013 22:42
|
|
|
Any of you guys able to log into Moodle for the Stanley class?
|
|
#
?
Oct 21, 2013 23:05
|
|
|
|
| # ? May 9, 2024 07:39 |
|
|
I'm just learning about the Stanley class so... see you guys summer of '14 I guess.
|
|
#
?
Oct 21, 2013 23:25
|
|
