theomega's questions

I want to configure varnish (still on version 3) in the following way:

If the backend is available, every request should go to the backend, so no cached results should be used. If the backend is unavailable (so not healthy), I want to use the cached results for the next day. If after one day the backend is still unhealthy, an error should be shown.

I tried the following configuration using the grace mechanism, but this doesn't work: Although all requests are going to the backend (so no cached answers if the backend is available), in the case of a failure, I instantly get an Varnish Error (Unavailable) instead of Varnish using the cached data. What am I doing wrong?

sub vcl_recv {
  if (req.backend.healthy) {
    # No caching if backend is healthy
    set req.grace = 1s;
  } else {
    # Otherwise accept requests up to 1 day old
    set req.grace = 1d;
  }
}

sub vcl_fetch {
  # Cached requests are not valid at all
  set beresp.ttl = 1s;

  # Keep requests for ttl + 1 d
  set beresp.grace = 1d;
}

Hy, I got a domain (say example.com) and a server (say IP 1.2.3.4). I want to relay every traffic which is requested from the domain to my server. Every subdomain should be handled like this (so wildcard DNS). There should be one exception: The subdomain mail.example.com should be routed to another ip (say IP 2.3.4.5)

My previous hoster made the following zone-file for the dns server for me:

@                        IN A       1.2.3.4
mail                     IN A       2.3.4.5
*                        IN CNAME   @

Now, my new DNS-Hoster claims this is invalid and proposes this new zone file to me:

@                        IN A       1.2.3.4
mail                     IN A       2.3.4.5
*                        IN A       1.2.3.4

Is the first one realy wrong? Is the second one correct and does what I want?

I already read I a Wildcard CName DNS Record Valid, so the first one should not be wrong but only unsuported.

Thanks!

Hy, I'm running Privoxy on my server and I only want to allow some users to use the proxy. I can't specify a IP-Address because they are dynamic. I also can't use ssh because those users use mobile devices where no ssh-client is available.

I know the Proxy-Protocol can use authenticated access to the server, does Privoxy support this?

Thanks!

Hy, a friend and me want to share a Linux-Machine. We both need to get root-rights via sudo for administering that machine. Is it somehow possible to deny the access to the home-folder for the other one, although he can become root?

Thanks!

Hy, I got several Linux-Servers (running Debian) where different services run: Database-Servers, Webservers, Applicationsservers, Tools and so on. All Servers are connected to the same internal network. There is also one special Server which is the Mail-Server: All Mailaccounts are stored on this server, it is also the outbound Mailserver for all the other servers. I want all Mails for all servers to get saved on the Mailserver. For example if an cron-job fails on one of the web-servers the mail should not be delivered to the local user but instead to the Mailserver so I get a centralized place for mail storage.

How do you set up this scenario?

My current setup is: Using postfix as MTA on the Mailserver and using ssmtp on all the other servers. SSMTP is configured to send the mails to the Mailserver. The Mailserver is configured to allow the whole internal network to relay mails using itself.

Is this the right way to choose? I also thought about setting up a MTA (postfix) on every server and configure it somehow to forward the mails. What would be the advantage of this solution?

I'm using libvirt and kvm to virtualize my server. I configured libvirt to start a dhcp-server on the bridge-network-interface to give the vm's their ips. Every VM's gets always the same ip based on its mac, this is configured like this:

<dhcp>
 <range start='10.1.1.2' end='10.1.1.254' />
 <host mac='54:52:00:21:01:ba' name='virstvm' ip='10.1.1.10' />
 <host mac='00:16:36:2d:71:f9' name='secvm' ip='10.1.1.20' />
</dhcp>

The problem: Whenever I add a new host to the dhcp section I have to restart libvirt-bin which restarts all my vm's. This cant be the solution because it means a downtime every time I add a new Server. Is there a solution?

Hy, I'm currently setting up a new Server using libvirt+kvm. Afterwards there should be about 5 virtual-machines on this server running (+some testing machines).

The storage is put on a raid-5 device which is set up using LVM. KVM now runs on some LVM-Logical Volumes.

The question is: Is there any drawback of using lvm again (a second time) inside the virtual machine to partition the space? So it would be: Harddisk-> Raid Controler -> LVM for Physical Server -> One Logical Volume per VM -> LVM inside each VM -> Several Logical Volumes inside each VM.

Are there any other possibilities if I want dynamic partitions inside my virtual machines?

Thanks

Hy, I got a brand new Intel i7 Server-System with 12GB Ram and I want to consolidate three other linux-servers onto this machine using virtualization. I want to use the harddisks (via a kind of LVM) in the server as storage as I got no SAN/IScsi. All Linux-VM's will run Debian-Linux, so no need for Windows-Support. I would prefer a bare-metal-virtualization (so where I dont have to maintain the os of "dom0"). The software should be free, even better open-source.

The virtual machines will run: PostgresSQL, a Lighttpd+PHP, a Tomcat-Server, a Mail-Server, a static-files Lighttpd.

The problem is, I cant decide which system to choose:

VMware EsXi: The Problem is the unsupported NIC in the server: It is a realtek-chip and as it is a hosted server I cant modifiy the hardware.

Citrix XenServer: I need the mangment Software to run under Linux, the Citrix-Software only runs on Windows. Bigger problem: You cant store the machines on the internal harddisk, you need a SAN.

XEN: Currently my favorite, but I found no managment software to use for showing the status of the virtual machines. I would have to setup a debian-system as dom0 and maintain it.

OpenVZ: I hate this, it never worked under debian right the way.

Main Problem is: the market is changing fast, so whatever I found on the internet was somehow outdated, so can anybody give me some advices? Thanks!

I maintain the servers for a big webproject (java + postgres + some tools around) which is currently hosted on three machines:

1. Machine: Mailserver (postfix), Ad-Server (lighttpd + php + openx)
2. Machine: Tomcat + Servlet
3. Machine: PostgreSQL-Server, static content (via lighttpd)

All machines run Debian Stable and are connected via a VPN (openvpn). As the hardware is very old (AMD Athlon 3000+ and 2GB RAM on each) its time for a change.

These servers should now be replaced by one big machine (16GB Ram, big Intel CPU's supporting VT, 5 IP's).

The question now is: Should I still seperate the differents tasks using Virtual Machines or should I simply put everything on the machine as usual. Where are the pros and cons?

I thought of the following:

Pro Virtualization:

• Security: As the vm's are seperated you cant take the whole machine (hopefully)

Con Virtualization:

• Performance: There is a performance loss
• Work: Every maintaince-work has to be done several times for every vm
• Communication: Communcation of the different vm (Servlet do Database) gets more complicated.
• Hard Memory Limits: I have to assign static resources (like Memory) to each machine. This can be a con if say my db-server needs more ram for 30 seconds (than it got assigned) and there would be more ram available on other machines. With no virtualization this wouldn't be a problem.

Thanks for any hints.