All-in-one homeserver, advisable?
-
I'd be surprised if there wasn't some virtualisation solution that would work for you. The overhead isn't that much for many cases.
What sort of bandwidth are we talking about here? Are you running packages or VPNs? It could be you're approaching the limits of what your Atom can do anyway but if not then try running as a VM.Steve
-
I'd be surprised if there wasn't some virtualisation solution that would work for you. The overhead isn't that much for many cases.
What sort of bandwidth are we talking about here? Are you running packages or VPNs? It could be you're approaching the limits of what your Atom can do anyway but if not then try running as a VM.Steve
WAN1: down: 2mbps up: 512kbps (ADSL)
WAN2: down: 7.2mbps (Mobile)
WAN3: down: 7.2mbps (Mobile)
LAN: Wifi n / 100mbps LAN.WAN1 is limited to 20 GB data transfer, after which it becomes 512mbps d/l. WAN1 is unreliable. After 5 GB of download WAN2 and WAN3 become unusable.
CPU usage on my Atom with pfsense and Asterisk is under 7%. I have 3 phones and <15 network devices (closer to 4 on average.)
-
So your maximum throughput is <20Mbps, assuming you're load balancing the three WANs. Even the slowest Atom would have no problems with that as you've seen. If you run Snort or VPNs though that won't be the case.
If you have the time try running as a VM. At the very least any performance results you get would be interesting.Steve
-
So your maximum throughput is <20Mbps, assuming you're load balancing the three WANs. Even the slowest Atom would have no problems with that as you've seen. If you run Snort or VPNs though that won't be the case.
If you have the time try running as a VM. At the very least any performance results you get would be interesting.Steve
My only choice here is to try virtualbox. My hardware doesn't support any type 1 hypervisor.
Again the ideal thing would be for each server to run in its own VM. That is 3 VMs, or at least 2: one for pfsense and one for Apache/Asterisk.
I don't think running Apache/Asterisk on the VM host would be smart. I'll get myself a lightweight linux os host and try to do this then. -
I don't think running Apache/Asterisk on the VM host would be smart. I'll get myself a lightweight linux os host and try to do this then.
Why??
CPU usage on my Atom with pfsense and Asterisk is under 7%. I have 3 phones and <15 network devices (closer to 4 on average.)
So this is doing NOTHING so why do you think you can not break it out to its own VM?? Your apache is not going to be a site like amazon now is it ;) So how much do you think it would draw??
-
I don't think running Apache/Asterisk on the VM host would be smart. I'll get myself a lightweight linux os host and try to do this then.
Why??
CPU usage on my Atom with pfsense and Asterisk is under 7%. I have 3 phones and <15 network devices (closer to 4 on average.)
So this is doing NOTHING so why do you think you can not break it out to its own VM?? Your apache is not going to be a site like amazon now is it ;) So how much do you think it would draw??
What I meant was that I'll have to run 2 VMs at least because Apache and Asterisk on a host that is running a VM with pfsense in it would be a potential security threat.
<rant>Quite honestly, maybe I am just biased, I can't imagine running three different OSes on that machine. I understand that virtualisation can give you near native performance now, I just don't think it can come anywhere close without 64-bit instructions, Vt-x and Vt-d.
I am saying this from previous experiences. I have a hard time keeping the machine responsive with just one OS on it.</rant>That said, I AM going to try running everything on vbox anyway. If everything works, great! otherwise I'll figure out a different solution.
-
here is my copy of 32bit ubuntu running on my esxi box – if its sitting there idle, its not going to be pulling any cpu, etc.
your other 2 vms are not going to be doing much of anything anyway - so just having the os installed and up an running is not really much overhead, etc.
-
I'd agree that an n2700 is fully capable of doing nothing when several VMs are loaded.
The problem is when all those VMs suddenly have something they should be doing at same time.
Thats when you will be hurting.
When I provision a system, I provision it with the idea of all the VMs working 100% under a full load.
I don't give any consideration to how well they work when idle.
-
Agreed he should be sized while loaded - but he has
"I have 3 phones and <15 network devices (closer to 4 on average.)" He is not serving up to the 100k users a website.. Pretty sure that cpu is going to set there idle.. With a few cycles here and there when he loads up a page via apache or a call comes in..
It sure is not going to be pushing packets with a 2mbps/512Kb connection ;)
-
http://www.ebay.com/itm/SUPERMICRO-1U-2x-L5420-Xeon-QC-2-5GHz-8GB-250gb-Half-Depth-Server-X7DCA-L-/141411937842
add ESXI and 8g more memory? 10 seconds on ebay.
-
Looks better.
-
So I did get a virtualbox to run on that machine with Ubuntu server 14.04 LTS. Apache/Asterisk VM was impossible to setup through virtualbox (it failed trying to configure the CD.) Managed to transfer a configured appliance and got everything running.
pfsense works just fine. Right now, the VM usage on host is 15% in top (effectively just 7.5%.) Although even small tasks spike the usage to 100%.
This is the same for the asterisk/tf VM. Something like package configuration make the CPU usage go crazy. And apart from that, passing through USB mobile broadband dongles seems to be such a pain (hasn't worked yet.)The only problems I am facing are, ZFS complaining of low memory on startup (potential problem.) High average loads (where ubuntu just refuses to give me load data at boot time.) and high CPU usage of the asterisk VM even on idle (around 40% I don't know if it'll scale well under load.)
What are the problems I am likely to face if I move apache and asterisk to the ubuntu host? I'll only open ports to apache.
Possible solutions that I thought of were to move apache to an OpenVZ container. Perhaps set up VPN access to apache on the VM host.
Apologies in advance. I realize that I am going offtopic now. I appreciate the help anyway.
PS: wcrowder: Used servers are a little hard to find where I live. ebay has few option. Other websites have used servers being sold in cities where there are more IT related companies. … and I am a college student, I'd need to save money for a few months before I could buy something worth $100. T_T
That said, I have back up options. An E3-1230v3 that sits idle (or is switched off) most of the time, and a C2D E4300 that's just lying somewhere doing nothing. I am just trying to find a use for this weak machine. -
"(it failed trying to configure the CD.)"
What?? Why would apache or Asterisk need a CD??
As to ZFS - that can be completely ignored not even used..
-
"(it failed trying to configure the CD.)"
What?? Why would apache or Asterisk need a CD??
As to ZFS - that can be completely ignored not even used..
I couldn't install Ubuntu. It gave me apt-cdrom errors and the set up crashed. I had to make an appliance and then use it there.
Setting up mobile broadband is giving me a headache. For some reason everything works fine for some time, and then my dongles disconnect because they aren't getting echo responses over ppp. After which the dongles refuse to connect. I don't know if it is because of the USB passthrough or just because pfsense doesn't support my modem.
-
Sure its not powering down the USB ports when you haven't touched the machine for a while?
-
Sure its not powering down the USB ports when you haven't touched the machine for a while?
Oh my God. Thank you! That worked. I wouldn't have thought that Ubuntu was powering the device down in less than a minute! It's been running for 7/8 minutes now.
EDIT: Scratch that. It failed eventually, just took a little longer than before.
Here's the log
Sep 27 15:11:22 ppp: [opt1_link0] LCP: no reply to 4 echo request(s) Sep 27 15:11:32 ppp: [opt1_link0] LCP: no reply to 5 echo request(s) Sep 27 15:11:32 ppp: [opt1_link0] LCP: peer not responding to echo requests Sep 27 15:11:32 ppp: [opt1_link0] LCP: state change Opened --> Stopping Sep 27 15:11:32 ppp: [opt1_link0] Link: Leave bundle "opt1" Sep 27 15:11:32 ppp: [opt1] Bundle: Status update: up 0 links, total bandwidth 9600 bps Sep 27 15:11:32 ppp: [opt1] IPCP: Close event Sep 27 15:11:32 ppp: [opt1] IPCP: state change Opened --> Closing Sep 27 15:11:32 ppp: [opt1] IPCP: SendTerminateReq #6 Sep 27 15:11:32 ppp: [opt1] IPCP: LayerDown Sep 27 15:11:34 ppp: [opt1] IFACE: Down event Sep 27 15:11:34 ppp: [opt1] IFACE: Rename interface ppp1 to ppp1 Sep 27 15:11:34 ppp: [opt1] IPV6CP: Close event Sep 27 15:11:34 ppp: [opt1] IPV6CP: state change Stopped --> Closed Sep 27 15:11:34 ppp: [opt1] IPCP: Down event Sep 27 15:11:34 ppp: [opt1] IPCP: LayerFinish Sep 27 15:11:34 ppp: [opt1] Bundle: No NCPs left. Closing links... Sep 27 15:11:34 ppp: [opt1] IPCP: state change Closing --> Initial Sep 27 15:11:34 ppp: [opt1] IPV6CP: Down event Sep 27 15:11:34 ppp: [opt1] IPV6CP: state change Closed --> Initial Sep 27 15:11:34 ppp: [opt1_link0] LCP: SendTerminateReq #2 Sep 27 15:11:34 ppp: [opt1_link0] LCP: LayerDown Sep 27 15:11:36 ppp: [opt1_link0] LCP: SendTerminateReq #3 Sep 27 15:11:38 ppp: [opt1_link0] LCP: state change Stopping --> Stopped Sep 27 15:11:38 ppp: [opt1_link0] LCP: LayerFinish Sep 27 15:19:07 ppp: Multi-link PPP daemon for FreeBSD Sep 27 15:19:07 ppp: Sep 27 15:19:07 ppp: process 90236 started, version 5.7 (root@pf2_1_1_i386.pfsense.org 08:21 15-Aug-2014) Sep 27 15:19:07 ppp: [opt1_link0] Link: DOWN event Sep 27 15:19:07 ppp: [opt1_link0] LCP: Down event Sep 27 15:19:07 ppp: [opt1_link0] LCP: state change Stopped --> Starting Sep 27 15:19:07 ppp: [opt1_link0] LCP: LayerStart Sep 27 15:19:07 ppp: [opt1_link0] Link: reconnection attempt 1 in 3 seconds Sep 27 15:19:07 ppp: caught fatal signal term Sep 27 15:19:07 ppp: [opt1] IFACE: Close event Sep 27 15:19:07 ppp: [opt1] IPCP: Close event Sep 27 15:19:07 ppp: [opt1] IPV6CP: Close event Sep 27 15:19:07 ppp: waiting for process 10039 to die... Sep 27 15:19:08 ppp: waiting for process 10039 to die... Sep 27 15:19:09 ppp: [opt1] Bundle: Shutdown Sep 27 15:19:09 ppp: [opt1_link0] Link: Shutdown Sep 27 15:19:09 ppp: process 10039 terminated Sep 27 15:19:09 ppp: web: web is not running Sep 27 15:19:09 ppp: [opt1] Bundle: Interface ng1 created Sep 27 15:19:09 ppp: [opt1_link0] Link: OPEN event Sep 27 15:19:09 ppp: [opt1_link0] LCP: Open event Sep 27 15:19:09 ppp: [opt1_link0] LCP: state change Initial --> Starting Sep 27 15:19:09 ppp: [opt1_link0] LCP: LayerStart Sep 27 15:19:23 ppp: [opt1_link0] CHAT: The modem is not responding to "AT" at ModemCmd: label. Sep 27 15:19:23 ppp: [opt1_link0] MODEM: chat script failed -
Are those plugged directly into the computer? Could you try it with a powered usb hub?
(running low on ideas)
-
Are those plugged directly into the computer? Could you try it with a powered usb hub?
(running low on ideas)
Don't have access to one right now. And I don't think that could be the problem (the dongle LED wouldn't be ON if it was powered down, right?) Anyway, I tried all the resources I found on setting the autosuspend mode off, nothing has worked so far. I checked the uptime the last time it failed, and that was about 16 minutes. The problem I am facing is similar to https://forum.pfsense.org/index.php?topic=45157.0
The dongle disconnects after sometime and it becomes impossible to reconnect it later (even removing and reinserting doesn't work.) When this happens, one of the PHP processes starts taking 40% CPU usage and must be killed manually for the system to work. It's just unreliable from that point onwards; I NEED to restart it to make the mobile internet or anything related to pfsense work.At the moment, I am using pfsense (64 bit) on VMware Workstation on a Windows 8.1 host and everything seems to be working perfectly. The link has been up for 40ish minutes and hasn't failed. I can remove the dongle as I wish, disconnect and reconnect without reboots.
Could it just be that the dongle doesn't work well with 32 bit pfsense? Maybe I should test 32 bit pfsense on VMware or perhaps just give up on this :/ -
It could be that the port just isn't pushing enough power on its on. I've experienced that many times with usb devices.
-
I would suspect some issue with the USB pass through to the VM. It's probably timing out somewhere. Maybe to save resources it only represents the USB device when it sees call to it. There could be many reasons. The USB dongle and pfSense really aren't designed/tested to operate in this way so you may be up against a insolvable issue.
Steve
