New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Linux Vserver, terrabytes of network traffic, or ifconfig garbeld?
I have my own little monitoring script, and I have added a few nodes to it, two of which are Linux Vserver based. Rest is OpenVZ or KVM. Now those two nodes have been up for 2 days, and my monitoring script tells me that the network interface has used hundreds of gigabytes:
|---[remy][vps12][~] |----> /sbin/ifconfig eth0 Link encap:Ethernet inet addr:127.0.0.1 Bcast:0.0.0.0 Mask:255.255.255.255 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:13686707336 errors:0 dropped:6322422 overruns:0 frame:0 TX packets:25873871936 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:2015520969817 (1.8 TiB) TX bytes:5085397843748 (4.6 TiB) Interrupt:30 Memory:f4000000-f4012800 lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 UP LOOPBACK RUNNING MTU:16436 Metric:1 RX packets:106690101 errors:0 dropped:0 overruns:0 frame:0 TX packets:106690101 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:60834318358 (56.6 GiB) TX bytes:60834318358 (56.6 GiB)
And the other one:
|---[remy][vps13][~] |----> /sbin/ifconfig eth0 Link encap:Ethernet inet addr:127.0.0.1 Bcast:0.0.0.0 Mask:255.255.255.255 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:1422925160 errors:0 dropped:1353207 overruns:0 frame:0 TX packets:1629653047 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:283762607040 (264.2 GiB) TX bytes:525456803652 (489.3 GiB) Interrupt:30 Memory:f4000000-f4012800 lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 UP LOOPBACK RUNNING MTU:16436 Metric:1 RX packets:144720979 errors:0 dropped:0 overruns:0 frame:0 TX packets:144720979 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:85762275126 (79.8 GiB) TX bytes:85762275126 (79.8 GiB)
Compare that with one of the OpenVZ nodes:
|---[remy][vps9][~] |----> /sbin/ifconfig lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 inet6 addr: ::1/128 Scope:Host UP LOOPBACK RUNNING MTU:16436 Metric:1 RX packets:502514 errors:0 dropped:0 overruns:0 frame:0 TX packets:502514 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:143152363 (136.5 MiB) TX bytes:143152363 (136.5 MiB) venet0 Link encap:UNSPEC HWaddr 00-00-00-00-00-00-00-00-00-00-00-00-00-00-00-00 inet addr:127.0.0.1 P-t-P:127.0.0.1 Bcast:0.0.0.0 Mask:255.255.255.255 inet6 addr: ::1/128 Scope:Host UP BROADCAST POINTOPOINT RUNNING NOARP MTU:1500 Metric:1 RX packets:31019381 errors:0 dropped:0 overruns:0 frame:0 TX packets:37963313 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:7019034930 (6.5 GiB) TX bytes:39501070496 (36.7 GiB)
Is this a Linux Vserver specific thing or is there something else wrong?
Comments
It's a vserver thing. You see stats for all traffic on the node.
No, vserver as in linux-vserver.
@sleddog thanks. That I find quite shitty, do you know why it has been done that way?
So it does not need to run through tr/iptables and slow the network speed down or create load.
For vserver explanation, maybe William is the best
Yeah i don't think there is anyone that still uses vserver except for Edis and Alvotech
Understand that, still don't like it for my own monitoring. Those two nodes are at Edis indeed.
@William why does edis uses Vserver? LXC for example is included in mainline kernel. Except for that Vserver is grownin in the software stack, if you could today start again, would you still choose for Vserver?
A bad choice when we started with it, can't change anymore
And no, we would likely use OpenVZ (But not with Solus, as we are based on Debian and not CentOS)
Why was it a bad choice? And, why did you start to use it?
Because it has more issues than advantages, no working IPv6 for example.
Someone else picked it.
Because it SUCKS MAJOR ASS
Vserver is such a complete joke, it's not even funny. Run as far as you can from it, even OpenVZ is worlds better than Vserver. And of course with EDIS you can choose to buy KVM instead.
These are limitations of the technology. If you need them then it's important, if you don't then it's irrelevant. OVZ has limitations too. For example it doesn't support ip broadcast (with venet). So you can't run a samba master or a dhcp server. But if you don't need that feature then it's irrelevant.
You also can't monitor the load of your VPS, as the load averages are those of the host node. On the other hand, vserver is the only virtualization technology that lets you monitor the host node's load from within the VPS
Which is easy to use and works fine. If you want a reactive firewall app on the VPS it's not going to work, but again, not everyone needs or wants that.
I bought a vserver from Edis a couple months back from a offer here. Less then $1/month. Been using it experimentally for web/mail/dns hosting, so those missing features don't matter for me. The load & transfer reporting is weird but I've gotten used to it.
During those two months the vserver has:
I've come to accept the Edis vserver for what it is -- something different that's damned reliable. I like it
I have my own firewall script uniform across all my VPSes (uploaded and updated automatically with some other scripts, only .conf changes a bit), why would I want to bother with some web-based firewall specifically on one particular host? And as I said the web firewall is host-specific, e.g. I only used Alvotech's, and it kind of sucked. Also while they were nice enough to patch in (experimental?) IPv6, there was no IPv6 support in the firewall. So while Alvotech similarly to your experience with EDIS was quite reliable and cheap, Vserver was so limiting and annoying, that I ended up cancelling it.
For my own hosting and virtualization needs proxmox and vagrant are good enough. And, I have both vserver and KVM over at Edis, and they work fine for what I do with them (lighttpd cluster). Just my monitoring on the Vserver nodes is messed up (http://vps11.sparklingclouds.nl/status/ vps12 and vps13.)
And for the firewalling, I can see how that is an issue. I only use firewall (iptables) on openVZ nodes, have not tried this yet. The Edis panel for firewalling is quite good...