|
Internet Explorer posted:So I know I asked about vShield in this thread before and got some good answers, but I was wondering if anyone has a specific AV product that works well for them. I need to get something in place and don't really have the time to do a ton of research. Anyone using something they like? Use vFlash with vShield and you won't realize any storage hit Mulloy posted:Question, I'm looking at a brief network pause (3-4 seconds at most) on a virtualized server. About 30 seconds prior to the stream of communication errors I see a VMTools update which includes a couple messages about Network Configuration updates occurring. Is it possible for an update like this to cause a brief pause/interruption in network activity? Are you using VMXnet? Are you using vDS? If VSS what is your fail-over/probing policy? Dilbert As FUCK fucked around with this message at 03:51 on Feb 25, 2015 |
#
?
Feb 25, 2015 03:41
|
|
|
|
| # ? Apr 18, 2024 02:56 |
|
|
Dilbert As gently caress posted:Use vFlash with vShield and you won't realize any storage hit I'm not worried about the storage hit. Just looking for advice on which vendor does a better job.
|
|
#
?
Feb 25, 2015 04:53
|
|
|
Internet Explorer posted:I'm not worried about the storage hit. Just looking for advice on which vendor does a better job. Trend probably has the best ratio for compute resources lost vs. infection detection. The UI/vApp was pretty poo poo but it got a lot better in recent releases.
|
|
#
?
Feb 25, 2015 05:02
|
|
|
Ran into an issue whereby during a power failure, had an issue with a ClusterSharedVolume volume name being changed for no good reason, has anyone run into this? Setup: 2 node 2012R2 cluster with attached MD3200i SAN over iscsi with MPIO. Clustered VM roles with VHDs stored on CSV volumes. After power came back and hosts/SAN were back up, trying to start a VM resulted in an error. Checked Application logs and noticed that the cluster thought the path to the VM disks/metadata was at C:\ClusterStorage\General-Disk-Volume Went to C:\ClusterStorage\ and noticed that this volume was somehow renamed to "volume1", e.g. C:\ClusterStorage\Volume1. I simply renamed volume1 back to its original name which I had given it, General-Disk-Volume, and VMs started up just fine. Any ideas on why windows renamed ONE of the volumes in there, but not the other? Bug in 2012R2? The order with which the SAN/Hosts came back online after power outage?
|
|
#
?
Feb 25, 2015 17:30
|
|
|
Anyone used veeam to backup to a remote quantastore?
|
|
#
?
Mar 4, 2015 05:18
|
|
|
We're a 100% Debian/Ubuntu shop and I've been tasked in trying out oVirt. There are no official packages for Debian and I really wanted to avoid installing RH or CentOS, so I tried building oVirt from scratch, following the documentation. It... didn't go well. Bunch of undocumented errors when running Maven tests and compiling, it was horrible. gently caress Java EE. I think I'll go with Xen or OpenVZ. We already use OpenStack but were not really satisfied and ran into strange problems (like sudden permissions errors when trying to access disk images - tracking that down amongst the logs of all the separate modules was a pain in the rear end). Senso fucked around with this message at 17:47 on Mar 6, 2015 |
|
#
?
Mar 6, 2015 17:44
|
|
|
Curious to see how the GPL vs VMWare lawsuit that was filed today goes, assuming the stuff in this FAQ is accurate, VMWare seems fairly shady. I also guess I didn't really realize how much of esxi was still basically just modified linux: http://sfconservancy.org/linux-compliance/vmware-lawsuit-faq.html
|
|
#
?
Mar 6, 2015 18:43
|
|
|
Senso posted:We're a 100% Debian/Ubuntu shop and I've been tasked in trying out oVirt. There are no official packages for Debian and I really wanted to avoid installing RH or CentOS, so I tried building oVirt from scratch, following the documentation. It... didn't go well. Bunch of undocumented errors when running Maven tests and compiling, it was horrible. gently caress Java EE. VDSM (which is basically the "glue" between libvirt and the ovirt engine) doesn't work on Debian-based systems as far as I know, which is almost certainly why you had a zillion problems. vdsm is python, but even had you made it past the engine, it would have been a bad time. This isn't a Java problem. It's a Debian problem. But oVirt is KVM, and "Xen" is not an alternative to oVirt unless you mean Xen CloudPlatform or XenServer, which are comparable to what oVirt actually does. OpenVZ is also not an alternative unless you intent to put HyperVM or whatever openvz frontend isn't dead on top of it. A community member put together Debian build instructions, but I've never tried them and can't vouch for them at all. Maneki Neko posted:Curious to see how the GPL vs VMWare lawsuit that was filed today goes, assuming the stuff in this FAQ is accurate, VMWare seems fairly shady. I also guess I didn't really realize how much of esxi was still basically just modified linux: They'll rewrite their driver shim to use a freebsd kernel to show that Linux isn't integral (even though they like to be able to re-use Linux drivers) and that their product isn't significantly based on the Linux kernel. evol262 fucked around with this message at 18:51 on Mar 6, 2015 |
|
#
?
Mar 6, 2015 18:48
|
|
|
evol262 posted:This isn't a Java problem. It's a Debian problem. But oVirt is KVM, and "Xen" is not an alternative to oVirt unless you mean Xen ButtPlatform or XenServer, which are comparable to what oVirt actually does. OpenVZ is also not an alternative unless you intent to put HyperVM or whatever openvz frontend isn't dead on top of it. Yeah I know Xen/OpenVZ are not the same. I'm not setting this up so we can have an interface to spawn and destroy instances on the go. We kinda wanted the whole cluster setup but meh, we'll see. Also, the Debian build instructions are useless, outdated and don't help.
|
|
#
?
Mar 6, 2015 18:54
|
|
|
Maneki Neko posted:Curious to see how the GPL vs VMWare lawsuit that was filed today goes, assuming the stuff in this FAQ is accurate, VMWare seems fairly shady. I also guess I didn't really realize how much of esxi was still basically just modified linux: Interesting, although its incredibly vague about what they are claiming vmware have done. I'm curious why they feel anyone cares about this bit as well. quote:Christoph is one of most active developers of the Linux kernel. He has contributed 279.653 lines of code to the latest Linux 3.19 kernel, and thus ranks 20th amongst the 1,340 developers involved in that release. Christoph also ranks 4th among those who have reviewed third-party source code, tirelessly corrected and commented on other developers' contributions.
|
|
#
?
Mar 6, 2015 18:58
|
|
|
jre posted:Interesting, although its incredibly vague about what they are claiming vmware have done. They're basically saying vmware took gplv2 code, incorporated it into their own vmkernel effort and then ignored the gplv2 license terms. quote:This case is specifically regarding a combined work that VMware allegedly created by combining their own code (“vmkernel”) with portions of Linux's code, which was licensed only under GPLv2 I'm assuming they're just giving some background on the guy because he's the one suing. Seemed reasonable to me. Maneki Neko fucked around with this message at 19:08 on Mar 6, 2015 |
|
#
?
Mar 6, 2015 19:05
|
|
|
Maneki Neko posted:They're basically saying vmware took gplv2 code, incorporated it into their own vmkernel effort and then ignored the gplv2 license terms. VMware's stance is basically that vmkapi is not derived code, since implementing an API which is compatible with GPL-licensed code is not a violation, and that vmkLinux is open, public code which abides by the GPL. VMware is probably in the right, honestly. I'm sure there'll be closed discovery and review, but VMware is denying that there's any Linux code at all in vmkernel itself.
|
|
#
?
Mar 6, 2015 19:20
|
|
|
Maneki Neko posted:They're basically saying vmware took gplv2 code, incorporated it into their own vmkernel effort and then ignored the gplv2 license terms. That's the vague bit. What do they mean by "incorporated" Linked against? copy and paste of GPL code ? quote:VMware's stance is basically that vmkapi is not derived code, since implementing an API which is compatible with GPL-licensed code is not a violation, and that vmkLinux is open, public code which abides by the GPL. VMware is probably in the right, honestly. Is this in any way similar to the Google vs Oracle case about google implementation of java APIs ?
|
|
#
?
Mar 6, 2015 20:02
|
|
|
jre posted:That's the vague bit. What do they mean by "incorporated" Linked against? copy and paste of GPL code ? Until there's discovery, it's hard to say. But it's likely that it's neither linked against nor copy and pasted. VMware basically said "it's hard for people to write device drivers, and some vendors would rather just re-use their existing work, so let's figure out a way to let them use drivers they've already written with as few changes as possible." So they took some of the driver subsystems out of the kernel and made them a modular piece called vmkLinux. The entire purpose of this piece of code is to let Linux drivers be used (relatively) natively, then it hooks into a shim called vmkapi which passes bits through to vmkernel itself. Probably DMA and all the other finicky bits drivers need, which should be trivial to re-implement. The problem with this, and why it's on shaky legal ground, is that this is obviously nowhere near a "clean room" implementation where they were handed compiled driver modules and told to let them interoperate with vmkernel. It's easy to look at the source for the kernel and see exactly what it's doing, exactly what the drivers are doing, and how all the pieces work. Then they can say "see, the GPL code never touches our proprietary code, because this API is the middle man!" Except how the hell would they have been able to write an API which almost perfectly meshes with the Linux kernel driver code from scratch? It's possible, but it's likely that they just looked at the code and figured that they'd be on safe ground for "derived" works. This is the question, if I'm understanding it right, and I may not be. jre posted:Is this in any way similar to the Google vs Oracle case about google implementation of java APIs ?
|
|
#
?
Mar 6, 2015 20:48
|
|
|
I just gotta say that I find it really funny that you probably understand this stuff better than 98% of people at VMware. Most people at VMware these days don't even know what a VM is.
|
|
#
?
Mar 7, 2015 02:32
|
|
|
DevNull posted:I just gotta say that I find it really funny that you probably understand this stuff better than 98% of people at VMware. Most people at VMware these days don't even know what a VM is.
|
|
#
?
Mar 7, 2015 08:01
|
|
|
Thanks for the explination, I didn't realise ESXi worked like that.
|
|
#
?
Mar 7, 2015 11:38
|
|
|
DevNull posted:I just gotta say that I find it really funny that you probably understand this stuff better than 98% of people at VMware. Most people at VMware these days don't even know what a VM is. I figured VMware probably tells employees not to comment, and you're the only person I'd expect to be able to give a real answer on how it fits together. jre posted:Thanks for the explination, I didn't realise ESXi worked like that. This is just the driver bits. They have their own scheduler and the rest of the pieces a kernel needs, as well as native drivers (the vmklinux stuff is legacy and has been since 5.1 or 5.5 or whenever the native driver toolkit emerged). It's heritage from RHEL being the esx management frontend.
|
|
#
?
Mar 7, 2015 15:10
|
|
|
evol262 posted:I figured VMware probably tells employees not to comment, and you're the only person I'd expect to be able to give a real answer on how it fits together. We got an email saying not to make comment on the lawsuit, but I don't think I know the vmkernel stuff well enough to make much more of a comment that what you described. I think you did a good enough job defending VMware anyhow. I guess it make sense for you to understand the competition and the general architecture of it. I generally try to stay away from ESX, and do 90% of my work on Workstation. It makes life a whole lot easier.
|
|
#
?
Mar 8, 2015 04:33
|
|
|
So I'm trying to assign a static IPv6 address to an Ubuntu VM on VMWare Player on a Windows 8 host so that is is globally reachable from any other computer but I can't quite succeed. At the moment, I've only configured a single network adapter for the VM, a NAT. /etc/network/interfaces is configured like this (following http://jochen.kirstaetter.name/blog/linux/configure-ipv6-on-ubuntu.html advice) : # The loopback network interface auto lo iface lo inet loopback # The primary network interface auto eth0 $ sudo nano /etc/network/interfaces auto eth0 # IPv4 configuration iface eth0 inet static address 192.168.1.2 network 192.168.1.0 netmask 255.255.255.0 broadcast 192.168.1.255 # IPv6 configuration iface eth0 inet6 static pre-up modprobe ipv6 address 2001:db8:bad:a55::2 netmask 64 However, after restarting, my eth0 interface is down. When I bring it up manually ifconfig only gives this: eth0 Link encap:Ethernet HWaddr 00:50:56:34:3f:24 inet6 addr: fe80::250:56ff:fe34:3f24/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:13 errors:0 dropped:0 overruns:0 frame:0 TX packets:5 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:1164 (1.1 KB) TX bytes:478 (478.0 B) I have no internet connectivity but the link address can be pinged from the host or other vms. So the goal is to have a VM that is globally routable from other machines (host, other vms, other computers...) with a well known IPv6 address that does not change. What am I doing wrong?
|
|
#
?
Mar 8, 2015 08:18
|
|
|
Kalenden posted:So I'm trying to assign a static IPv6 address to an Ubuntu VM on VMWare Player on a Windows 8 host so that is is globally reachable from any other computer but I can't quite succeed. Using nat ? Surely you want a bridged interface if the vm is to be externally reachable on its own ip address?
|
|
#
?
Mar 8, 2015 13:05
|
|
|
Kalenden posted:
Use a bridged NIC instead, and don't bother with static IPv6 addresses, if you have working v6 with a router sending RAs your autoconfig IPv6 address will be 'static' anyway since it's based on your MAC.
|
|
#
?
Mar 8, 2015 14:43
|
|
|
DevNull posted:We got an email saying not to make comment on the lawsuit, but I don't think I know the vmkernel stuff well enough to make much more of a comment that what you described. I think you did a good enough job defending VMware anyhow. I guess it make sense for you to understand the competition and the general architecture of it. I generally try to stay away from ESX, and do 90% of my work on Workstation. It makes life a whole lot easier. Once upon a time I was a pretty good vmware guy who knew the architecture really in depth. The rest of my team is hilariously unaware of VMware and what they offer, as I discover every time I work the booth at a conference and they're totally unable to relate our features to yours or answer questions like whether or not we have anything like vSomething. There's almost no emphasis on understanding the competition.
|
|
#
?
Mar 8, 2015 16:13
|
|
|
evol262 posted:The rest of my team is hilariously unaware of VMware and what they offer, as I discover every time I work the booth at a conference and they're totally unable to relate our features to yours or answer questions like whether or not we have anything like vSomething. There's almost no emphasis on understanding the competition.
|
|
#
?
Mar 8, 2015 16:58
|
|
|
Not knowing any details about your competitors offerings is more common than it should be.
|
|
#
?
Mar 8, 2015 17:13
|
|
|
adorai posted:Is that because they don't consider VMware to be competition? Or do they just plain not care? It's because they're developers who have never worked as admins/engineers dealing with it. And we don't consider VMware competition. Product management certainly does, but what VMware is or isn't doing doesn't really matter to us when we're working, individually. "Gluster and oVirt hyperconvergence" certainly is a analogous to vSAN. Whether the guys doing the implementation know that is another question. How VMware does it doesn't matter to us at all, in any way, since we're using totally different tools and we couldn't re-use their architecture even if we wanted to, so why learn it?
|
|
#
?
Mar 8, 2015 17:14
|
|
|
Does anyone have strong opinions on MariaDB/MySQL vs. PostgreSQL as a backend database for OpenStack? I've never used the MySQL replication options and would feel a little more comfortable with PostgreSQL replication. We're looking at a fairly small deployment of < 10,000 instances with a small handful of images and very few users.
|
|
#
?
Mar 8, 2015 19:41
|
|
|
Misogynist posted:Does anyone have strong opinions on MariaDB/MySQL vs. PostgreSQL as a backend database for OpenStack? I've never used the MySQL replication options and would feel a little more comfortable with PostgreSQL replication. We're looking at a fairly small deployment of < 10,000 instances with a small handful of images and very few users. The database doesn't see a lot of traffic either way. I use postgres (though I'm never above ~500 instances in my environment) even though the official recommendation is mysql, I think
|
|
#
?
Mar 8, 2015 19:56
|
|
|
We're using Percona XtraDB Cluster for OpenStack (which as you probably know is a souped up version of MySQL and the Galera multi-master clustering tech). No issues with it at all. All of our DBA expertise is in MySQL so we didn't bother evaluating anything else once it proved to work fine.
|
|
#
?
Mar 8, 2015 21:16
|
|
|
evol262 posted:It's because they're developers who have never worked as admins/engineers dealing with it. And we don't consider VMware competition. Product management certainly does, but what VMware is or isn't doing doesn't really matter to us when we're working, individually. Because the terminology is useful, if people come from a MS shop they will ask about Active Directory like features. It is easier if your people know that they are talking about centralized management of users and some SSO stuff instead of "No, we don't do that." It makes it easier for customers if you can compare stuff by its name. IMHO.
|
|
#
?
Mar 9, 2015 08:33
|
|
|
Double post?
|
|
#
?
Mar 9, 2015 08:33
|
|
|
Mr Shiny Pants posted:Because the terminology is useful, if people come from a MS shop they will ask about Active Directory like features. It is easier if your people know that they are talking about centralized management of users and some SSO stuff instead of "No, we don't do that." In a specific sense, we don't interact with customers. Sales does. They can probably compare stuff by name. Or they could when I was external and they were trying to pitch RHEV to the VMware shop I worked at. But we only interact with customers working the booth for upstream projects at conferences, or through support cases. The support cases obviously do not use VMware's marketing names. The people you talk to at conferences are either technically capable enough to be able to describe vWhatever in non-marketing terms or managerial players who don't actually know what features VMware has either. Upstream feature planning, RFEs (even customer-filed RFEs), and everything else goes the same way. I've never seen a comparison to VMware marketing terms in an engineering context. And again, we don't interact with customers, so that analogy is useless. In a very specific sense, I don't know what vmkapi and vmkLinux are because I want to understand the competition. How they're doing it doesn't matter to me technically. If I wanted to create a vShield analog, I'd have to do it from scratch using technologies fundamentally different enough from vmkernel that their methodology would be useless (if I worked on libvirt or KVM directly, maybe I'd be able to cross-apply). I know it from being a VMware guy before I came here, but I wouldn't know vShield details enough to give a good architecture overview either. In general, I think this sentiment is some feel good thing goons tell each other about working in business. But it's totally meaningless whenever you start talking about engineering companies and not small IT departments in non-IT organizations who get requests to re-implement AD for free and they need to know that it's LDAP+krb+DNS+dhcp+cldap plus client-side hooks. But we would never see that. If some customer were to ask for that, they'd do it through a TAM, who'd put it in generic terms with specific requirements, and the assigned engineer would write something which fulfills those. You will never talk to anyone in my group unless you go ask at one of the upstream booths at a conference. Usually an open-source conference. Customers will never talk to anyone in my group. From a purely Red Hat perspective, many of our engineers (most, probably) are more passionate about the upstream work and providing a viable open alternative than we are about specific business use cases. To pick on vShield again, it's a "nice to have", in theory. But we're way more interested in something like Cinder integration, because that has a lot of community interest. 1:1 mappings to other people's features don't matter to us. Again, product management knows and cares. But engineering doesn't, shouldn't have to, and it doesn't make them better. It may make them worse, since knowledge of a feature gap just makes you want to reach parity, but we'll just play catch up forever if we do that. From a high-level business sense, it's nice to know what other people are doing (even if most of VMware's addons, using VMware as an example, don't get used by most of their customers).But it requires also having a very broad overview of what a bunch of different teams in your own organization are doing, and relating their tasks. This is long winded and only related to this thread because we're talking about VMware and I work in the cloud group building virt products for Red Hat, but this "knowing your competitor's marketing terminology is useful" stuff is pointless at best and harmful at worst for many development and engineering roles.
|
|
#
?
Mar 9, 2015 15:07
|
|
|
Could I get some suggestions for monitoring software/solutions? VMware/Microsoft environment, two offices, ~25 vm's. Looking to monitor basically application availability - so... windows services monitoring mainly. Also, anyone aware of a way to get an email alert based off of iscsi latency? ie. "Datastore has had 100ms or higher for 5 minutes." Using ESXi 5.1 at the moment and the only alarms appear to be connectivity related, not performance related.
|
|
#
?
Mar 9, 2015 15:08
|
|
|
Trying to make some upgrade decisions here. We have a 3 node VMWare 5.1 cluster right now. We are going to be adding 3 more hosts to it in the next 2 months (already have the hardware, just trying to plan out the upgrades.) Right now, we don't really have enough capacity to put a host in maintenance mode (one of the reasons why we are adding more hosts) which is the main reason why we are still on 5.1. So, I'm trying to figure out the best way to move forward. Add in these 3 hosts at the same VMWare version as our current hosts and then upgrade them one at a time after the fact? Add in the new 3 hosts at 5.5 or 6.0 and then upgrade the old ones one at a time? Establish a new cluster with the new hosts and migrate the old hosts one at a time into the new cluster as they are being upgraded?
|
|
#
?
Mar 9, 2015 15:16
|
|
|
Here is the official response from VMare on the lawsuit for anyone interested: http://www.vmware.com/company/news/vmware-update-to-mr-hellwigs-legal-proceedings.html
|
|
#
?
Mar 9, 2015 15:21
|
|
|
Anyone ever seen "error 6" in Failover clustering logs? 2012R2 failover cluster. I get the below when I run a Get-ClusterLog and output to a file: 2015/03/09-18:49:38.466 ERR Error 6 Been googling but having a hard time with this - Starting up my 2012R2 box (1 of 2 nodes in a failover cluster), the issue is that the Cluster Service will time out and not start upon boot, but WILL start when I manually start the service and runs seemingly fine. The service is set to the default, which is Automatic start. I tried changing the timeout value on the service to no avail. Ran all available windows updates. No AV or anything weird on the server. I'm using CSV with attached iSCSI storage using a Dell MPIO driver, I'm wondering if this is somehow related to the delay that windows is taking to get the iscsi sessions set up during boot and how it starts the Cluster Service, but I'm just kind of guessing in the dark a bit here. Event logs don't show anything useful that I can see under System logs other than the service failed to start after the standard 30000ms timeout.
|
|
#
?
Mar 10, 2015 00:16
|
|
|
bull3964 posted:So, I'm trying to figure out the best way to move forward. As long as the hardware matches, you can bring it in at 5.5 and migrate/update the 5.1 hosts. If they are different models (mainly CPU differences) it gets more complicated. Also make sure vCenter is at least 5.5 prior to updating, obviously - I have not tried mixing 5.1 / 6.0 hosts in clusters so I can't say to that version.
|
|
#
?
Mar 10, 2015 07:17
|
|
|
ragzilla posted:Use a bridged NIC instead, and don't bother with static IPv6 addresses, if you have working v6 with a router sending RAs your autoconfig IPv6 address will be 'static' anyway since it's based on your MAC. Figured out I need a bridged one. Problem is, I need a static IP(v6) address because my host and thus the VM will be constantly moving from and to different networks (not all of which auto-provide me with a IPv6 address) and using a link-local autoconfig address is insufficient. Edit: it is best I'm more complete so: I want to have a Ubuntu VM running on VMWare Player on a Windows 8.1 Host that is globally reachable/addressable by a static/non-changing ip address. By other VMs, by other hosts, by other types of devices, etc... . The host machine will move between LANs often, sometimes to a wired setup sometimes to a wireless setup. The VM has, at the moment, a bridged adapter to both the wireless and ethernet network adapter. I I have tried to configure both dhcp conf of VMWare Player as the VM itself but either I lose all connectivity, the statically chosen assigned IP doesn't stick or something like that happens. Also a difficulty is that I'm moving between networks and gateways often. Can anybody give a foolproof description on how to do this? To reiterate: a Ubuntu VM that moves between networks a lot and that I can reach from anywhere. Preferably with an IPv6 address. Kalenden fucked around with this message at 10:28 on Mar 11, 2015 |
|
#
?
Mar 11, 2015 09:23
|
|
|
Kalenden posted:Can anybody give a foolproof description on how to do this? To reiterate: a Ubuntu VM that moves between networks a lot and that I can reach from anywhere. Preferably with an IPv6 address. When you say 'from anywhere', do you mean that no matter what layer 2 network you're attached to you'll be able to access the VM from other hosts in that layer 2 network? Or do you mean any host on the Internet should be able to access the VM despite where it's powered on?
|
|
#
?
Mar 11, 2015 12:49
|
|
|
|
| # ? Apr 18, 2024 02:56 |
|
|
ragzilla posted:When you say 'from anywhere', do you mean that no matter what layer 2 network you're attached to you'll be able to access the VM from other hosts in that layer 2 network? Or do you mean any host on the Internet should be able to access the VM despite where it's powered on? The second. Preferably, devices not on the same network should be able to reach the VM.
|
|
#
?
Mar 11, 2015 16:42
|
|