|
Basically everything's a webserver nowadays. I'm going to start discovering the IP of everything around me.
|
#
?
Apr 11, 2014 20:13
|
|
|
|
| # ? Apr 23, 2024 19:52 |
|
|
Roargasm posted:I need some help figuring out if I'm way in over my head or just retarded. Why not both? Engage someone to help you, or request training. Please don't wing it with storage.
|
|
#
?
Apr 11, 2014 20:24
|
|
|
evol262 posted:Log into the management interface, and you'll be presented with options (I only have no idea what the interface on this is; HP is big on Lefthand, so maybe that, but I've never touched this particular bit of kit) Having used HP MSA 2000 G3's before, it does have a GUI. It's Hitler distilled down into GUI form, but it's a GUI 
|
|
#
?
Apr 11, 2014 20:42
|
|
|
three posted:Why not both?
|
|
#
?
Apr 11, 2014 22:22
|
|
|
TheFace posted:Got a random question and figured here is as good a place as any. I have a ESXi 4.1 host that is acting strange. The only alert is about unavailable space, which I know means I need to get into the console and clear out some logs (or find space somehow). Problem is SSH won't let me in (even though it was running, I stopped it when it didn't work and tried to start it again and it failed to start), and the console isn't letting me in either. I can't telnet to port 902 either (which would normally allow but tell me ssl required and not let me do anything). At the console press alt f1 and it will give you a terminal to run commands
|
|
#
?
Apr 11, 2014 22:41
|
|
|
Thanks guys, I'll find a class to take or professional to mentor me before I dig into this further
|
|
#
?
Apr 11, 2014 23:09
|
|
|
Roargasm posted:Thanks guys, I'll find a class to take or professional to mentor me before I dig into this further This is a good book to get you started with storage. It is from EMC, but not EMC-specific. https://education.emc.com/ISMbook/default.aspx
|
|
#
?
Apr 14, 2014 21:19
|
|
|
Turns out HP's new Gen8 IvyBridge implementation is buggy as all hell - I've got 16Gb Windows VMs BSODing if you use the hardware vMMU and random hosts PSODing if you have more than 256Gb ram installed with the factory BIOS. It's been a fun week...
|
|
#
?
Apr 17, 2014 11:50
|
|
|
Mausi posted:Turns out HP's new Gen8 IvyBridge implementation is buggy as all hell - I've got 16Gb Windows VMs BSODing if you use the hardware vMMU and random hosts PSODing if you have more than 256Gb ram installed with the factory BIOS. What version of ESXi are you running? We've got some Gen8 hosts out at a remote site (South Korea, I'm in Aus) and you're making me worried 
|
|
#
?
Apr 17, 2014 15:16
|
|
|
Mausi posted:Turns out HP's new Gen8 IvyBridge implementation is buggy as all hell - I've got 16Gb Windows VMs BSODing if you use the hardware vMMU and random hosts PSODing if you have more than 256Gb ram installed with the factory BIOS. Just a tip for anyone running HP servers. Its almost always the firmware/bios. Upgrade it anytime you have problems. Save yourself the headache. Even in instances where the hardware has been static and running okay for a long period of time. It just doesn't make sense.
|
|
#
?
Apr 17, 2014 15:22
|
|
|
cheese-cube posted:What version of ESXi are you running? We've got some Gen8 hosts out at a remote site (South Korea, I'm in Aus) and you're making me worried Seems to specifically apply to the E5 26xx v2 CPUs (v1's we are running seem ok), but generally we are having issues with the Ivy Bridge chipset coming out of the factory at the moment. Sickening posted:Just a tip for anyone running HP servers. Its almost always the firmware/bios. However the BSODs with large VMs when using the hardware vMMU is not fixed by the latest SPP, in fact they haven't even publicly acknowledged it yet that I've seen - probably because they haven't figured out which vendor is to blame yet.
|
|
#
?
Apr 17, 2014 15:34
|
|
|
I'm trying to get two different teams working in two different virtualization platforms (Citrix and OpenStack) to work together and nobody seems to know anything about what the other is doing. So I come to you for my question: Is there a simple way to convert a citrix vm-export (xva) to something that openstack can use (vmdk or qcow2)? I just need to be able to load my exports from Citrix into OpenStack. I'm assuming this should be simple, but that is probably my first mistake.
|
|
#
?
Apr 17, 2014 18:17
|
|
|
I only have experience with the OpenStack half, but they both claim support for OVF containers and VHD disks. http://docs.openstack.org/image-guide/content/image-formats.html http://support.citrix.com/proddocs/topic/xencenter-62/xs-xc-vms-export-ovf.html
|
|
#
?
Apr 17, 2014 18:32
|
|
|
Yeah, I saw there was some overlap in the supported formats. That's why I was hoping there was some utility I could use to easily convert into something they could both use. I know just enough about virtualization to add resources or spin up a new image, but my day to day is far removed from infrastructure. I was just hoping there was an easy way to go about doing this that would be painless for everyone involved. If at all possible I want to find a simple way to convert between formats without doing more exports. If I can just take care of it myself then neither of the engineers gets to complain about me giving them "more work."
|
|
#
?
Apr 17, 2014 18:39
|
|
|
HatfulOfHollow posted:I'm trying to get two different teams working in two different virtualization platforms (Citrix and OpenStack) to work together and nobody seems to know anything about what the other is doing. So I come to you for my question: Is there a simple way to convert a citrix vm-export (xva) to something that openstack can use (vmdk or qcow2)? Just use XenConvert. But this isn't how Openstack works. You can import a qcow, but Openstack is not XenServer or VMware. Those teams are not doing the same thing, and consolidation is not necessarily a good path forward. You can convert it and push it into glance, then configure a machine to boot from that volume, and it'll basically work. But don't put pets on Openstack. HatfulOfHollow posted:Yeah, I saw there was some overlap in the supported formats. That's why I was hoping there was some utility I could use to easily convert into something they could both use. I know just enough about virtualization to add resources or spin up a new image, but my day to day is far removed from infrastructure. I was just hoping there was an easy way to go about doing this that would be painless for everyone involved. If at all possible I want to find a simple way to convert between formats without doing more exports. If I can just take care of it myself then neither of the engineers gets to complain about me giving them "more work." Don't take this the wrong way, but virtualization != virtualization. If you know "just enough" to add resources or spin up a new image, you should investigate the use cases for Openstack and cloud computing in general, watch this, figure out whether the team using Openstack should be using something entirely different (if they're all pets, Openstack is not a good use case -- oVirt will let them use the Neutron and Glance providers they already have while they migrate), etc. If they're both pets -> consolidate on XenServer If they're both cattle -> consolidate on Openstack If they're mixed -> use xenapi as a nova-compute backend if you have the management and team go-ahead to consolidate If they're mixed -> do not shove XenServer VMs into Openstack evol262 fucked around with this message at 18:46 on Apr 17, 2014 |
|
#
?
Apr 17, 2014 18:41
|
|
|
The thing is... I don't need to know how to do it. I just need to know if it's possible. We've got two disparate projects that were started by independent teams and when they became aware of each other it made sense to try to re-use some of the things I've been working on. My project is building out a new data center. I've got xva exports of images that will be used to spin up new integration and QA environments as they are needed. The other project is run by an entirely different team, but they're using openstack to stand up developer sandboxes on demand. So the end result is very similar. The thing is that I've automated nearly the entire deployment process but it requires the exports with some software preloaded on them if they want to use the utilities I've already developed. So that's why I'm trying to load my xva into their openstack. The alternate route is to setup another image for openstack the same way as mine and just maintain both, but I'd prefer to just avoid that divergent path if at all possible. Consolidating on one platform is not going to happen. There are two different divisions within the company and two different architects running these projects and they both have egos the size of houses. deedee megadoodoo fucked around with this message at 19:21 on Apr 17, 2014 |
|
#
?
Apr 17, 2014 19:10
|
|
|
HatfulOfHollow posted:The thing is... I don't need to know how to do it. I just need to know if it's possible. We've got two disparate projects that were started by independent teams and when they became aware of each other it made sense to try to re-use some of the things I've been working on. My project is building out a new data center. I've got xva exports of images that will be used to spin up new integration and QA environments as they are needed. The other project is run by an entirely different team, but they're using openstack to stand up developer sandboxes on demand. So the end result is very similar. The thing is that I've automated nearly the entire deployment process but it requires the exports with some software preloaded on them if they want to use the utilities I've already developed. So that's why I'm trying to load my xva into their openstack. The alternate route is to setup another image for openstack the same way as mine and just maintain both, but I'd prefer to just avoid that divergent path if at all possible.
|
|
#
?
Apr 17, 2014 19:38
|
|
|
evol262 posted:Is this Linux, or a sysprepped Windows image? All RedHat.
|
|
#
?
Apr 17, 2014 19:39
|
|
|
HatfulOfHollow posted:All RedHat. Then you don't need your tools. Everything you need to do can and should be done with a combination of kickstarts and configuration management. Every manual step you're doing, put it in a kickstart. Lorax/ImageFactory will build images which can be consumed by both XenServer and Openstack from a single script. Basically, Openstack images need to do the following: Disable avahi Shrink the root filesystem as much as possible Trigger an expansion of that filesystem when an image comes up (so it'll work properly across multiple flavors) Inject SSH keys and scripts through cloud-init (which is why avahi needs to be off) Ideally, not use password auth and disable root login Be as generic as possible. Openstack does not do any of these things for you. cloud-init does some. Go read these. If you aren't using configuration management (puppet, chef, whatever), why not? Failing that, if you can't do your configuration/customization in a kickstart, why not? Pulling random images into Openstack to boot from isn't remotely close to best practice or even good practice. Do you fulfill these requirements? If so, dump it into XenConvert and pull the image into Glance. If not, here be dragons, and you need to re-evaluate how you're doing things before you think about running it on openstack.
|
|
#
?
Apr 17, 2014 19:57
|
|
|
So I finally have the chance to deploy an (Essentials licensed  ) vCenter instance in our datacenter in a production facing role.We are going to be putting some smaller customer facing web services on these machines (HA/Failover will be accomplished with a LB and two VMs for now). We aren't going to be using shared storage either (no SAN for now). I'm planning on a spec with 1TB of disk space per VM host (4 disks RAID-10) w/2 smaller SSDs in RAID-1 for any faster VMs we require/swp file storage. My questions are 1) should I bother with SSD's for swap storage and 2) should I go with 15k RPM drives for my datstore or are slower (but larger) 10k RPM SAS disks alright for this? Like I said these are going to be webservers that will ideally be running many off-host DB queries, so the disk load is likely to be quite light. Wicaeed fucked around with this message at 20:13 on Apr 17, 2014 |
|
#
?
Apr 17, 2014 20:09
|
|
|
So yeah helping people build reference architectures is a challenge, especially when the person you are up against is a vcdx/emc engineer. A lot of the things he is doing is making me sratch my head in a wtf manner and I'm probably wrong I guess but I don't know; been a bunch of things I worked with him a year or so ago and got shot down because "that's not how you should do things" only to wind up full circle on a "yeah that thing and what not you wanted to do that's how we should do it". Probably going to be venting a lot here or in the lab thread don't know which one is more suitable, for VM/Net/Storage/'vBlock' designing.
|
|
#
?
Apr 19, 2014 02:14
|
|
|
What is a realistic minimum memory scenario for VCSA in a lab environment? I'll only be managing one ESXi host with maybe a half-dozen VMs. I just want VCSA so I can do things like creating templates, etc. I haven't picked up a new system yet so this is running on a C2D with 8 gigs RAM so obviously I want to keep resources minimal. I don't mind bringing VCSA down when I'm not using it so I guess I COULD give it a fair amount of resources, but ehh.
|
|
#
?
Apr 20, 2014 02:37
|
|
|
Martytoof posted:What is a realistic minimum memory scenario for VCSA in a lab environment? I'll only be managing one ESXi host with maybe a half-dozen VMs. 4Gb min, 2 vCPU's. 6GB is ideal.
|
|
#
?
Apr 20, 2014 02:46
|
|
|
DAF have you been drinking again?
|
|
#
?
Apr 20, 2014 11:31
|
|
|
Dilbert As gently caress posted:4Gb min, 2 vCPU's. 6GB is ideal.  :[
|
|
#
?
Apr 20, 2014 13:21
|
|
|
If your hosts are utilizing a bunch of ram you may want to look into forcing small pages at the 4K sector.
|
|
#
?
Apr 21, 2014 19:36
|
|
|
Hmm, that is likely one thing I'll have to do. Thanks for the advice  Is there a good go-to document on making this happen? some kinda jackal fucked around with this message at 01:01 on Apr 22, 2014 |
|
#
?
Apr 22, 2014 00:57
|
|
|
Dilbert As gently caress posted:If your hosts are utilizing a bunch of ram you may want to look into forcing small pages at the 4K sector.
|
|
#
?
Apr 22, 2014 01:17
|
|
|
As a totally ghetto workaround to not having VCSA installed I just said gently caress-it and exported my 2012R2 base image as an OVF template. I'll mess with VCSA when I have a machine with more than 8 gigs. Speaking of which, I just inherited 32GB of DDR2 ECC in the form of a hollowed out 1U system. Thinking about dropping my specs a little and picking up something that'll make use of that. Maybe an older Xeon system?
|
|
#
?
Apr 22, 2014 01:21
|
|
|
Martytoof posted:As a totally ghetto workaround to not having VCSA installed I just said gently caress-it and exported my 2012R2 base image as an OVF template. I'll mess with VCSA when I have a machine with more than 8 gigs. It may or may not actually work with 5.x - need to check the HCL to confirm.
|
|
#
?
Apr 22, 2014 01:28
|
|
|
Daylen Drazzi posted:It may or may not actually work with 5.x - need to check the HCL to confirm. Oh we've got some old poo poo running 5.5u1 at work. I'm really not too concerned with finding something that will work
|
|
#
?
Apr 22, 2014 01:38
|
|
|
We have an old Win 98 box that runs a production machine via a parallel port. Management wants me to virtualized it since the hardware is 15 years old and could die at any moment. ...uh, where do I begin here? I'm no stranger to virtualization, but the parallel port part is throwing a monkey wrench into things.
|
|
#
?
Apr 22, 2014 01:45
|
|
|
You want a serial to IP converter. Something like this http://www.lantronix.com/device-networking/external-device-servers/uds1100.html
|
|
#
?
Apr 22, 2014 02:02
|
|
|
adorai posted:It does it automatically at a certain threshold. Forcing it is not necessary. Yeah when you hit at like 94% mem utilization. http://www.boche.net/blog/index.php/2013/03/19/large-memory-pages-and-shrinking-consolidation-ratios/ it actually does quite a bit for you when you force 4k. Martytoof posted:Hmm, that is likely one thing I'll have to do. Thanks for the advice It's Host>config>Advanced settings and change "Mem.AllocGuestLargePage" to 0 2MB pages are good for dynamic memory servers and java; but in many environments is not noticable. One you enable it you may need to wait about 30 minutes, or reboot the host. CPU may seem a small bit higher but for the ram savings.... Dilbert As FUCK fucked around with this message at 03:41 on Apr 22, 2014 |
|
#
?
Apr 22, 2014 03:30
|
|
|
Internet Explorer posted:You want a serial to IP converter. Something like this Would that work for parallel output too? i.e. LPT ports?
|
|
#
?
Apr 22, 2014 13:45
|
|
|
Parallel to Ethernet stuff tends to be targeted at printers. Whether that will work depends on your application. Are you sure it isn't 25 pin serial?
|
|
#
?
Apr 22, 2014 14:20
|
|
|
ThinkFear posted:Parallel to Ethernet stuff tends to be targeted at printers. Whether that will work depends on your application. Are you sure it isn't 25 pin serial? Looking at the program now trying to figure that out. It's some DOS-based proprietary software that just runs as a .EXE file, with no real configuration options  But after talking to some of the crew back there, the serial to ethernet device looks like it will work.
|
|
#
?
Apr 22, 2014 15:51
|
|
|
If it's just some DOS thing then maybe I'll get laughed out of the building for suggesting this, but try replacing the Win98 portion with DOSBox. We did that with one of our big clients and replaced a Win98 machine running a single EXE in a DOS window with a DOSBox instance running on a CentOS VM and it does a fantastic job. They've got a poo poo ton of arcade games talking to it and it hasn't broken down in something like the five years since it was put in place. I think DOSBox has some parallel port passthrough options as well. *I'm really only mentioning this to get Windows out of the equation, not to dissuade you from going the P2Ethernet route or anything.
|
|
#
?
Apr 22, 2014 18:11
|
|
|
I'm looking at upgrading my VMware cluster from Xeon 55xx's to 56xx's so I can jump from 4 to 6 cores per socket, since CPU load is getting up to the 60-70% range during peak. The current processors support 1066mhz dimms while the 56xx ones support 1333mhz. We don't exactly run a huge workload, small-ish db's and a handful of burst cpu load GIS servers along with all the normal AD/Exchange and assorted other low-load crap. I'm not sure if there is even a way to see available memory bandwidth to tell if its worth swapping out the dimms or not. My gut says no, but I don't have a way to verify aside from doing the upgrade and seeing if performance stagnates as I deploy more VM's. We're at about a 4.25 vCPU/core ratio right now if that helps.
|
|
#
?
Apr 22, 2014 19:00
|
|
|
|
| # ? Apr 23, 2024 19:52 |
|
|
For a 266MHZ bump for what you will pay is most likely not going to see any noticeable increase in performance. http://www.vmware.com/files/pdf/techpaper/VMW-vSphere-Performance-FAQ.pdf (top right) You'd probably see better performance boosts out of leaving the 56xx/55xx generations.
|
|
#
?
Apr 22, 2014 19:09
|
|