Anthony's questions

Not sure if I should asking this here but am quite stuck at the moment.

I wouldn't be the best on Linux but can get around. My issue is with Yum update on my CentOS 6.3 x64 system.

Basically for some reason after installing Murmur (Mumble) it won't let me update - I have a feeling it is to do with one or more of the required packages.

This is what I get when running 'yum update':

yum update
Loaded plugins: fastestmirror, priorities
Loading mirror speeds from cached hostfile
 * base: ftp.heanet.ie
 * epel: ftp.heanet.ie
 * extras: ftp.heanet.ie
 * updates: centos.weepeetelecom.nl
Setting up Update Process
Resolving Dependencies
--> Running transaction check
---> Package bash-completion.noarch 1:1.3-5.el6 will be updated
---> Package bash-completion.noarch 1:1.3-7.el6 will be an update
---> Package db4-devel.x86_64 0:4.7.25-18.el6_4 will be obsoleted
--> Processing Dependency: db4-devel for package: 4:perl-devel-5.10.1-136.el6.x86_64
Package db4-devel is obsoleted by db53-devel, but obsoleting package does not provide for requirements
---> Package db53-devel.x86_64 0:5.3.21-1ice.el6 will be obsoleting
---> Package mod_python.x86_64 0:3.3.1-14.el6.1 will be updated
---> Package mod_python.x86_64 0:3.3.1-16.el6 will be an update
--> Finished Dependency Resolution
Error: Package: 4:perl-devel-5.10.1-136.el6.x86_64 (@base)
               Requires: db4-devel
           Removing: db4-devel-4.7.25-18.el6_4.x86_64 (@base)
               db4-devel = 4.7.25-18.el6_4
           Obsoleted By: db53-devel-5.3.21-1ice.el6.x86_64 (zeroc-ice)
               Not found
 You could try using --skip-broken to work around the problem
 You could try running: rpm -Va --nofiles --nodigest

Now I can see two familiar items, based on the install instructions for Murmur:

1 - perl-devel-5.10.1-136.el6.x86_64: I think this relates to epel-release-6-8.noarch.rpm

2 - db53-devel-5.3.21-1ice.el6.x86_64: Which possibly relates to this, zeroc-ice-el6.repo

Would I be on the right track here? If so how would I go about resolving this?

The install instructions are here:

http://mumble.sourceforge.net/Install_CentOS6

• Under the heading: New packages 1.2.4 (x64)

Any help is appreciated, even if I am only pointed in the right direction.

Thanks all.

UPDATE >>

So I tried 'Michael Hampton' solution and all was fine until installing murmur; all I get is the following:

yum install murmur
Loaded plugins: fastestmirror, priorities
Loading mirror speeds from cached hostfile
 * base: ftp.heanet.ie
 * extras: ftp.heanet.ie
 * updates: ftp.heanet.ie
Setting up Install Process
No package murmur available.

So then I tried to just install it, and got this:

rpm -ivh https://www.dropbox.com/s/hs0ztzrdy4unn75/mumble-server-1.2.4-1.el6.i686.rpm
Retrieving https://www.dropbox.com/s/hs0ztzrdy4unn75/mumble-server-1.2.4-1.el6.i686.rpm
error: Failed dependencies:
    libIce.so.35 is needed by mumble-server-1.2.4-1.el6.i686
    libIceUtil.so.35 is needed by mumble-server-1.2.4-1.el6.i686
    libQtCore.so.4 is needed by mumble-server-1.2.4-1.el6.i686
    libQtDBus.so.4 is needed by mumble-server-1.2.4-1.el6.i686
    libQtNetwork.so.4 is needed by mumble-server-1.2.4-1.el6.i686
    libQtSql.so.4 is needed by mumble-server-1.2.4-1.el6.i686
    libQtXml.so.4 is needed by mumble-server-1.2.4-1.el6.i686
    libcap.so.2 is needed by mumble-server-1.2.4-1.el6.i686
    libcrypto.so.10 is needed by mumble-server-1.2.4-1.el6.i686
    libdns_sd.so.1 is needed by mumble-server-1.2.4-1.el6.i686
    libgcc_s.so.1 is needed by mumble-server-1.2.4-1.el6.i686
    libgcc_s.so.1(GCC_3.0) is needed by mumble-server-1.2.4-1.el6.i686
    libgcc_s.so.1(GLIBC_2.0) is needed by mumble-server-1.2.4-1.el6.i686
    libprotobuf.so.6 is needed by mumble-server-1.2.4-1.el6.i686
    libssl.so.10 is needed by mumble-server-1.2.4-1.el6.i686
    libstdc++.so.6 is needed by mumble-server-1.2.4-1.el6.i686
    libstdc++.so.6(CXXABI_1.3) is needed by mumble-server-1.2.4-1.el6.i686
    libstdc++.so.6(GLIBCXX_3.4) is needed by mumble-server-1.2.4-1.el6.i686
    libz.so.1 is needed by mumble-server-1.2.4-1.el6.i686
    qt-sqlite is needed by mumble-server-1.2.4-1.el6.i686

Lastly I tried:

yum --nogpgcheck localinstall mumble-server-1.2.4-1.el6.i686.rpm

On the package itself and the same complaint.

Is it that the murmur install, for YUM, is not linked up on CentOS? It appears that it is Fedora. Sorry I don't know the correct terminology here.

As per the question: "Under What Conditions May PHP Server Variables Be Disabled".

Basically on Shared Hosting I am using I can choose to run my website in 'Apache' or 'CGI' mode. When running in 'CGI' mode the PHP Server Variable 'https' is not available, it simply does not exist where in 'Apache' mode it does.

Would this be a result of Safe Mode? Or something similar? Or would this have to be explicitly disabled in the config of PHP?

That's about it, I have had a lot of trouble trying to find an answer for this.

I am not asking for how they may have their hosting setup, I am asking WHY would certain PHP Variables be disabled, or in this case specifically the HTTPS variable, and would it be as a result of Manual interaction, PHP config, or part of some other 'system' i.e. PHP Safe Mode. Is the HTTPS variable a security risk?

I'm looking to setup a Spam Filter Proxy for my clients, not too many mind you, and have it as transparent as possible.

I read about SAVASM and Spam Vigilante, and while they are perfect I guess for the job, I can't install them directly to the intended Cloud VM.

I figure Spam Assasin is the way to go and would like to deploy it myself, or learn how to, as it will be on a Linux Cloud VM.

The OS it will be going onto is CentOS - does anyone have any better suggestions as to waht OS flavour they think this should go on?

Also does anyone know of any well rounded resources or how to guides to go about this?

While I know it isn't recommended to block 'bad' email outright, unless there is some kind of web portal I can have the clients log in to to check blocked email, I would have to have the system store the 'bad' email for x many days should the clients come looking for it.

Also is it possible, with Spam Assassin, to forward on to different SMTP servers depending on the mail domain it is intended for?

Thanks in advance.

Ok so I looked at a number of resources, per Google, and questions on here an none seem to be what I require.

I come from a Windows IIS background so maybe my thought process is flawed and that Apache does things differently.

Basically on IIS it is recommended to set up a dedicated application pool per web user, or website even, and have their files/folders stored under their own path and work with that dedicated application pool.

I am setting up Apache for testing, and learning purposes, and came across the situation of segregating users. Does Apache need to be forked per user or is the 'www-data' sufficient?

Seeing as the 'separate' webdirs are in separate 'home' directories I assume this is safe? My understanding is that if you can't access a directory somewhere in the path, anything behind that can't be accessed - is this correct?

Lastly, I noticed, and is the reason why I am asking this, that Wordpress when making files, or cache files, is creating them as user 'www-data' and group 'www-data'. This is what prompted me to ask as I am assuming now that the 'user' doesn't have access to those files now?

I have looked into this error and it seems that it hasn't been discussed yet - or at least I can't find any information relating.

I'm having issues transferring files, usually larger files over a couple of hundred MB.

Here is the setup:

• QNAP 410 as iSCSI Target with multiple LUNs. (CRC is turned on (Data Digest and Header Digest)
• Server 2003 with iSCSI Initiator version 2.08 - build 3825

(I'm copying files from anothe machine to shares on Server 2003 => into TrueCrypt volume ergo onto the NAS)

I have mounted the LUN and formatted it with TrueCrypt using NTFS (Full format, not a quick one).

What happens is some files, mainly RAR/Compressed files, appear as if they copy but fail. I've tested this in a number of ways and can repeat the process every time.

So I thought to check transfer over iSCSI without TrueCrypt in between, a plain NTFS format - no problem at all.

So it would seem TrueCrypt is at least part of the problem here.

I haven't tried copying directly from the server yet, I will try that. I also haven't tried it without CRC but fail to see how that would affect this. I will update with my findings later.

In the meantime does anyone have any ideas as to what could be wrong?

Thanks for your time.

Update:

I copied a set of files, the ones I was having issues with, to the server then from there I copied those into two places within the TrueCrypt volume (Mounted on the NAS).

1. A seperate directory create in the root of the volume
2. The same initial directory I was using in the first instance

Both worked fine. So it now seems clear that this is a link between TrueCrypt, iSCSI and Windows Shares.

I say this because I originally setup the whole system using TrueCrypt volume files, not iSCSI. I changed it as it didn't suit my requirements - day wasted as well. While I had this setup though I copied my entire file set to the volume files and all files copied without error - over the network, from a pc, to the server where TrueCrypt had the volume files mounted from the NAS.

I didn't bother turning off CRC on the iSCSI system as I highly doubt that is the cause in light of this finding.

So any ideas?

I've noticed that my web server, 2008 Xen VM, gradually loosing free space - more than I would of though from normal use and decided to investigate.

There are two problem areas:

*C:\Users\Administrator\ (6,755.0 MB)*

with files: 

NTUSER.DAT{randomness}.TMContainer'0000 randomness'.regtrans-ms
NTUSER.DAT{randomness}.TM.blf

AND

C:\Users\Administrator\AppData\Local\Microsoft\Windows\ (6,743.8 MB)

with files

UsrClass.dat{randomness}.TMContainer'0000 randomness'.regtrans-ms
UsrClass.dat{randomness}.TM.blf

From what I understand these are in-time backups of registry changes. If that is the case I cannot possibly understand why there would be 10000+ changes. (That's how many files there are per folder location, over 20,000 per folder in total.)

The files are using almost 15GB of space and I want rid of them, I'm just wondering can I remove them. However, I need to understand why they are being created so I can avoid this in the future.

Any ideas why there would be so many? Is there a way I can check to see what is making the modifications?

• Are they created with login attempts?
• Are they created in relation to every day Web Server use?
• etc. and so on

This is a strange one, even for me.

I have a number of mailboxes setup on Exchange 2003 which I access from the same Outlook instance. Oddly only one has this issue.

Basically when I go to do an operation on one or more emails in the mailbox, on a delete it will happen in every case, for some reason it seems like there is a bottleneck and the connection to the server is stalled/lost until Exchange catches up to what it is that I want to do.

It's odd as it does not happen with the other mailboxes, which are setup in the exact same way, and any operation on those is almost instant.

Basically, this mailbox is the 'support@' mailbox and I simply delete mails once I have noted, read them.

Any ideas why this could be happening?

Thanks for your time.

I've tried to resolve this myself and have looked through Google and Stack but haven't found the answer I'm looking for.

Currently on a VPS server I have BIND DNS installed as a MASTER DNS Server. I use Zerigo's DNS service as SLAVE servers for public use: The Master doesn't receive queries - It's job is to simply create and modify DNS entries locally of which the SLAVE use to serve.

Here is an excerpt of the BIND log, I set it to INFO event logging:

14-Apr-2012 23:00:00.234 general: info: received control channel command 'reload'
14-Apr-2012 23:00:00.234 general: info: loading configuration from 'C:\DNS\BIND\etc\named.conf'
14-Apr-2012 23:00:00.234 general: info: using default UDP/IPv4 port range: [1024, 65535]
14-Apr-2012 23:00:00.234 general: info: using default UDP/IPv6 port range: [1024, 65535]
14-Apr-2012 23:00:00.250 general: info: reloading configuration succeeded
14-Apr-2012 23:00:00.250 general: info: reloading zones succeeded
14-Apr-2012 23:16:22.750 xfer-out: info: client 174.36.24.251#47135: transfer of 'ajmakeup.com/IN': AXFR started
14-Apr-2012 23:16:22.750 xfer-out: info: client 174.36.24.251#47135: transfer of 'ajmakeup.com/IN': AXFR ended
14-Apr-2012 23:16:23.015 xfer-out: info: client 68.71.141.22#36212: transfer of 'ajmakeup.com/IN': AXFR started
14-Apr-2012 23:16:23.031 xfer-out: info: client 68.71.141.22#36212: transfer of 'ajmakeup.com/IN': AXFR ended

As you can see there is no problem with Zerigo's DNS servers requesting new DNS data, when I force a reload that is; I don't believe, as per the way they are set as SLAVE, that they poll for changes.

However the problem is the other way; the MASTER is not updating the SLAVE servers when reload is run (on the MASTER); it is a batch on a 15 minute timer.

Below is my NAMED.CONF:

key "rndc-key" {
    algorithm hmac-md5;
    secret "REMOVED FOR SECURITY";
};

acl "trusted" {
        174.36.24.251/32;
    68.71.141.22/32;
        localhost;
};

options {
    version "not currently available";
    directory "C:\DNS\BIND\etc";
    allow-query {
                trusted;
        };
};

controls {
    inet 127.0.0.1 port 953
    allow { 127.0.0.1; } 
    keys { "rndc-key"; };
};

logging{
  channel simple_log {
    file "C:\DNS\BIND\logging\bind.log" versions 3 size 5m;
    severity info;
    print-time yes;
    print-severity yes;
    print-category yes;
  };
  category default{
    simple_log;
  };
};

zone "ajmakeup.com" in {
    type master;
    file "c:\dns\BIND\zones\db.ajmakeup.com.txt";
    allow-transfer { 174.36.24.251; 68.71.141.22; };
    allow-update { none; };
};

Does my problem have something to do with 'allow-query' under options? You will notice that 'allow-transfer' is set explicitly on each DNS zone.

In case you need it here is my RNDC.CONF:

key "rndc-key" {
    algorithm hmac-md5;
    secret "REMOVED FOR SECURITY";
};

options {
    default-key "rndc-key";
    default-server 127.0.0.1;
    default-port 953;
};

server localhost {
  key "rndc-key";
};

Note:

I am using WebsitePanel as my hosting panel and is such why it creates the zone enteries the way it does. Although I know I can change this behaviour, I do not wish to do so nor do I believe is the root of the problem.

Thanks for your help.

I wasn't sure which was best to post in so I posted here and on http://webmasters.stackexchange.com

I will delete the other if this post is more relevant here.

I'm having a slight issue with memory issues on my web server. It is Windows 2008 RC2 IIS 7.5 with Fast CGI installed and running, all up to date.

I have a number of Wordpress websites running on this server, yet they do not all run at the same time; server load/traffic is not that high at the moment.

However when some of them run, even if it just me on the site, they spawn multiple php-cgi.exe instances to handle my "surfing".

Some points to note:

• I have Hyper Cache installed on each Wordpress site, and seems to be working fine.
• I understand that the PHP-CGI.exe instances only stay alive for as long as required.

For example: one Wordpress website with pretty much no content and minimal plugins (Hyper Cache and Google Analytics) spawn:

• The w3wp.exe instance and levels out around 10MB
• 5 x php-cgi.exe which are each roughly 9MB

That amounts to 55-60MB at first run. It seems a bit extreme to me. Another site is pretty much the same at first boot but spawns 8 x php-cgi.exe instances. To put it another way, I would expect this usage from ASP.NET not PHP.

Can I limit the php-cgi.exe to one instance per website and have it handle all requests for that website?

Another point I should mention, I am running PHP 5.2.17. If I upgrade to PHP 5.3.* will this problem go away? (Not that won't break some sites but I can live with that)

Would my only solution to be move them to a Linux box or will I have the same issues doing so?

Thanks for your time and help.

Best I give a back story so you get the full picture.

I am in the process of setting up a new business, which hopefully will start in April 2012, and as such I am preparing every aspect of it. One part of the business will be web hosting.

I am quite experienced in this area and have managed VPS before. In my case, I went for two servers to start with; the first is just a web server, the second has mail and databases.

While MS SQL databases won't be available initially to the customers it may or may not be added with an "upgrade" etc. either way I have 2008 express installed as I require it for Website Panel (which I may or may not end up using). Otherwise MySQL databases are free to use for customers.

I was pondering the backup strategy for both database systems.

For the MS SQL I have one daily full backup at 00:00 and the differentials on every hour from 01:00 to 23:00. I figured for now this would be the best way to backup the "control panel" - I may end up doing log backups every 10 mins also when I go live.

As for MYSQL, it seems I can't do differentials, like I did with MS SQL, where the preferred action is to do a full dump each time. I'm sure you can imagine disk space will fill up quite quick over time in this respect.

My main question is: how should I be backing up MySQL hourly?

One full backup daily is not good enough for customers, no matter if it is in writing or not. I do realize in reality customers are responsible for their data and backups and indeed this is how most companies work but I want to add the extra safety net that may or may not help - best effort if you will.

I read somewhere that replicating MySQL to another machine then doing hourly dumps would solve my purpose but then I would be worried about the lag. I have read that replication an cause bottlenecks. Is this still the case?

Ideally, I would like to have another server for staging/backups/replication but this is not in the budget right now. Maybe when I get a certain number of customers I will do this.

Anyway, that's pretty much the bulk of it.

Thanks for your time and help.

P.S

I thought to mention, just in case it is not obvious, I am using Windows Server for bot VPS - In this case 2008 R2 Data Center.

Also, in addition to my backups - I ship them off over the net to iDrive as online backup. My idea, which works perfectly, is that I create the backup files with the exact same names as previous. This way, on the online backup, I only use up the size of data that is the local backup files yet while keeping a 30 day "previous version" set online.

So I can say to my customers I have 30 (min) days worth of backups, which I do in reality all while officially only using up one backup set space.

Possible Duplicate:
Can you help me with my software licensing question?

So simple enough question really, I hope. I'm a little confused by Microsoft's licencing, go figure eh.

I'm maybe looking at providing Hyper-V virtual machines for rental, that is hosting them for rent by customers. Much like you'd get your own VPS from a provider.

Well, I realise with my model I should go with Windows 2008 DataCentre Edition licences, that's the easy bit. My question is do I require External Connector licences (EC), for external users?

If I do, is it per physical machine or per virtual machine/container?

It's very hard to determine this from Microsoft's published documents. They say if the customer/partner is accessing/authenticating to your network then for each server they can access an EC licence is required. Does this though apply to only active directory systems or?

Basically, if I setup all customer's VMs as workgroup servers, just like a VPS I rent, is then a EC licence required. In a nut shell, yes they may be accessing the server/network where the VPS is but they are not specifically autheticating against servers. Essentially this VM is "theirs".

Simple enough really. Can I create a IPSEC rule/filter between two publically available servers (they are actually on the same network) while still allow external traffic to the same resources.

I'll try to give as much info as possible:

Web Server <--> Database Server, of which both have Public IPs.

I want to secure database traffic/connections from the web server but yet still allow external direct connections, from my office, to the database server, if needed.

The thing is, and the reason I ask all this, is that the two servers are Virtuzzo servers in a data centre somewhere, maybe even on the both hardware node. I'd like for the "between" server traffic to be secure at least.

I found many step by step guides on how to setup the filters, I just want to confirm my request.

I am currently renting a VPS from a service provider in the UK. They supply Parallels Virtuozzo VPS. My particular issue is that I cannot use "Shadow Copies" at all on this virtual machine/container. Which in turn means I cannot use Open-File backup from any backup client, including ntBackup/Windows Server backup. Hence my main issue.

The spec: Windows 2008 RC2 64bit Data Centre, C: part of 100GB, 2GB RAM. HDD and RAM is barely used at the moment.

So, I have done quite a bit of research online and have come to the conclusion that the fault is with the Virtuozzo container. Here are my observations:

• Shadow Copies and related Open file backups work fine on Hyper-V, VMWare and standalone hardware installs. As such I have tested the extact same backup routine on each.

• I have tried the same routines on three other Virtuozzo containers, two 2008 server and one 2003 server. All have the exact same issue. Note that this was three seperate suppliers. So it stands to reason Virtuozzo is the culprit.

• I've gone through the removal and re-registering of relevant DLLs. No joy.

• I tired to use VSSADMIN to manually create references and shadowstorage but no joy.

• Most resources I have looked up refer to Windows Hotfixes or Windows updates. My hands are tied with both of these, as are most customers, on Virtuozzo containers as the supplier controls the updating.

• A few resources even point to disabling the "Parallels VSS Writer" in the client machine. Google "r1soft continuous data protection". Look for "Continuous Data Protection Enterprise Edition" and click the "VMs" tab. Under the heading "Virtuozzo for Windows" gives proof to this. I have been told not to do this as it will affect the VPS operation. I'm not convinced it will affect it to be honest but can't chance it, yet I don't know how to disable it anyway.

• Linked Here. This individual may have the answer yet I can't try the fix as I am unsure how the VPS would react. My reckoning is that when creating and deploying a Virtuozzo template a similar set of events happen and the newly created VE is looking for the host's shadow copy data or similar.

• Linked Here Parallels refers to the same issue but makes the point that shadow copies are not allowed on the C: partition. I don't believe this should be the case. I mean: on nearly every, shadow copy enabled, server out there, there is a Shadow Copy storage on the C: partition by default. Heck its even on my Windows 7 machine out of the box. Why do Parallels get to decide where it is stored. Maybe others can clarify why? If it is even relevant to my issue.

Without sounding too peeved, Shadow Copies should be a standard allowed practice as without it we can't backup open files. This to me seems to be a serious flaw.

I have asked the suport section of my provider to get Parallels involved but not sure how that will pan out. I have had to do this as Parallels only offer direct support to licensed users or partners. I would have to pay a support fee otherwise. Yer right, not for a fundamental flaw like this.

While I would like a straight answer from Parallels, that may take a while, I thought I might ask the community if anyone else has had similar issues.

Thanks in advance for help towards this, it is very much appreciated.