Stann's questions

Amazon has two toolsets that have to be used for ami creation ec2 api tools and ec2 ami tools which both require EC2_HOME environmental variable to be set. So it seems like I can only use one of them toolsets at a time in a single terminal? and if I have to use another toolset - I have to export EC2_HOME variable to correct location every time I need to witch toolset - is this correct?

What would one do if I want to use both toolsets from single terminal. Iknow that I can open multiple terminals and setting EC2_HOME to various location that way - but it seems like a bit contr-productive.

any ideas?

I use Fedora as my home machine and thus wanted to use Centos 6.0 linux as amazon EC2 instance. Here is the thing though - Amazon gives you long list of very unscientifically sorted instances, some of them as I understand Centos 6.0:

158864314994/mesos-hypertable-centos-6 944964708905/rightimage_centos_6.0_i386_20110810.1_ebs 944964708905/rightimage_centos_6.0_86x64_20110810.1_ebs 452552107804/Bayscribe-SearchServer2-CentOS6-EBS_9-8 452552107804/Bayscribe-SearchServer2-CentOS6-EBS_9-9

or something like this.

What do I have to do to get my beloved standard minimal netinstall of Centos 6.0?

Does Amazon maintains official images? because how do I know if these 3-rd party images don't contain some malicious crap?

I run Ubuntu Linux and use backup script that copies things over using rsync:

rsync -rc /home/user/source /media/nas/destination

or

rsync -r /home/user/source /media/nas/destination

However - it takes approximately 2.5 hours every day (i run this every day), to get it done. There is approximately 76 Gb of data over 1Gbit lan. So - basically - rsync just doesn't skip existing files in destination - it just copies everything over and over again.

I assume that the problem lies in NAS (d-link 321) and that it's a samba share that doesn't support unix persmission system - right? So this way rsync can't distinguish old files from new files - right? I've tried rsync with -c parametr (skip based on checksum, not mod-time & size) but still no dice, still takes 2.5 hours.

Can anybody suggest the way of syncing appx 70-80 Gb of data that doesn't ivolve "dumb" copying of everything every time?

Thanks:)

UPDATE

well. no luck. I've erased destination directory and and ran this twice:

rsync -r --times /home/user/source /media/nas/destination

It still took 2.5 hours both times. so it copies everything twice.

mmm. I went deeper to investigate this problem. And it looks like rsync doesn't preserve times. I've ran stat command on some files from source and destination - and Access, Modify, Change times aren't preserved on destination. I have no clue as to why does this happen... Any other suggestions? P.s. I relatively new to Linux - sio it can be something really dumb and silly:)

UPDATE 2 Resolved: problem as I realized stemmed from me mounting shares incorrectly. I've been mounting them like this (in /etc/fstab):

//192.168.10.199/Volume_1 /media/Volume_1 cifs   guest,rw,iocharset=utf8,file_mode=0777,dir_mode=0777 0 0

And this is what worked:

//192.168.10.199/Volume_2 /media/Volume_2 cifs    
username=sam,password=some_password,uid=developer,gid=developer,_netdev 0 0

Rsync preserves times now and only takes 3 minutes!

I'm using Ubuntu server right now (for local development) and noticed that some Ubuntu packages like apache have their config files... um... scattered.

While official Apache docs primarily refer to httpd.conf for everything, Ubuntu's Apache packages cleverly divides this one file into ..um...>20 files.

For example, instead of adding virtual hosts in httpd.conf you are expected to create files in a subdirectory, then alias it to another directory...m'ok. I guess many people like this...but I'm..um...think this overcomplicates things. if I want my config separated - I can do it myself.

What Linux server distro in your opinion has their packages customized the least by distro creators?

I see that quite often other people's web-site directories have empty html index files (1 file per dir). AFAIK - that's been done to prevent website users from seeing directory content - right?

I usually disable indexing in httpd.conf differently:

<Directory />
   Order Deny,Allow
   Deny from all
   Options None
   AllowOverride None
</Directory>

What would be the reason to create index.html in every web faced directory instead of simply restricting indexing in apache config file?

I've installed CentOS 5.5 with apache, php and mysql.

Right now the /var/www/html directory is owned by the root while apache is being run by apache user from apache group.

Is that recommended? Should I chown the html directory to another user such as wwwuser:wwwgroup?

After default CentOs 5.5 install - Document_Root directory (/var/www/html) is owned by root. In my httpd.conf file I have:

user apache group apache

mmm. Should I change ownership of /var/www/html to apache:apache?

UPDATE

Never mind. I was trying to figure out why I can't access a directory that i've created inside of the Document_Root. Then I've stoped SELinux and was able to access that directory.

I know that the cornerstone of mysql backup is mysqldump which works for all engines. I also see that some folks advocate the use of LVM snapshots.

What would be the point at which you stop using mysqldump and start creating LVM snapshots?

Is mysqldump more suitable for smaller databases and LVM snapshots - larger ones?

Thanks!

I've ran into problems trying to backup data from table City in world database from mysql 5.0 server on local network:

mysqldump --tab=/home/developer/Desktop -u developer -h 192.168.10.196 -p world City

I'm getting this error:

Not dumping tablespaces as no INFORMATION_SCHEMA.FILES table on this server

My mysqldump version is 10.13 (from Mysql server 5.1.49) While server on 192.168.10.196 has Mysql 5.0.77 installed (default mysqldump version that comes with that package is 10.11).

Does mysqldump version makes a difference? Why can't I use mysqldump?

Thanks!

I've been using mysql for development. I don't use replication but I guess I had binary-logging enabled for some time on one of my previous computers and I have unknowingly accumulated hundreds of binary log files.

I'm using Linux now and I actually need to start using binary logging. When I try to enable binary log in my.cnf - mysql server doesn't start and this is what I find in error log:

  1 110314 14:39:40 [Note] Plugin 'FEDERATED' is disabled.                                                                                                                                                                                                                        
  2 110314 14:39:40  InnoDB: Started; log sequence number 0 36669358
  3 ^G/usr/sbin/mysqld: File '.\mysql-bin.000179' not found (Errcode: 2)
  4 110314 14:39:40 [ERROR] Failed to open log (file '.\mysql-bin.000179', errno 2)
  5 110314 14:39:40 [ERROR] Could not open log file
  6 110314 14:39:40 [ERROR] Can't init tc log
  7 110314 14:39:40 [ERROR] Aborting
  8 
  9 110314 14:39:40  InnoDB: Starting shutdown.

so - I assume - I just have to delete old binary logs - right? Can I just delete all mysql-bin. files? Would I have to clear mysql-bin.index file as well?

Thanks!

just curious if there any tool that I can just give a mail server address/ip to and that would give me all features that mail server supports.

Like: does it support imap/pop3? Does it support tls/ssl/ntlm encryption? etc....

thanks:)

so... I've been wondering lately why do I need GDM. I got it disabled for experiment sake through modifying upstart file /etc/init/gdm (I run Ubuntu 10.10 desktop). So now computer boots to command prompt and I just type in startx if I need GUI.

so far everything runs just fine. But - does anyone know if there are any drawbacks to not using gdm?

Would I lose any functionality?

Thanks:)

I'm new to Linux. just studying php/mysql development. On windows I kept Apache, PHP, Mysql, PgSQL directories in one folder on SEPARATE DRIVE, like this:

D:\server
    \mysql
    \apache
    \php

So - If I would to reinstall Windows - all I would have to do is start mysql and httpd services - and that's that. quick and easy. The same like XAMPP, except I don't need 20 extra apps that xammp give you.

I would love to do the same in Ubuntu. So I would keep apache,mysql,php on separate drive and start them whenever I want to (init.d or upstart to start them automatically).

so - the structure would be:

/media/storage/server
/media/storage/server/mysql
/media/storage/server/apache
/media/storage/server/php

I got everything compiled and setup aside from config files.

The problem is:

I want all config files to be in one place as well. I mean http.conf -in apache folder. my.cnf - in mysql folder

so - Question: How do I make MySQL load configuration file not from /etc/my.cnf, but from another arbitrary location. couldn't find this info anywhere..:(

important:

1)it's not a production server! It's just for someone who studyis web development and linux and doesn't want to keep reinstalling and reconfiguring server when he changes distros. 2) I know it's against Linux hierarchical standards. 3) I now about Ubuntu's tasksel lamp-server and/or XAMPP easinness:)

I want to manually install Apache server. I've compiled it into:

/server/apache

http.conf configured correctly. It works. I can open up a browser and navigate to localhost and I can see "It works message".

But how does one goes about adding apache into Ubuntu's startup so I won't have to do:

sudo /server/apache/bin/apachectl start

all the time?

Can somebody explain how does one goes into adding programs to Ubuntu startup (10.10 64 bit)?

RESOLUTION

I learned a lot. It looks like there are 2 ways of doing it:

1) Historical ways: most Unix/Linix distros historically followed System.V initialization patter. This way all u have to do is write a special start/stop/restart script, chmod +x it and put in under /etc/init.d directory. Then you run update-rc.d command which creates links under different runlevels. and that's how it works. la la la

2) New way. Many linux distros currently switching from old runlevel based system to event based initialization. In my case Ubuntu (also RHEL 6.0 and Fedora) uses system called Upstart which eventually will completely replace systemV version. it uses /etc/init folder. All you have to do is create a script, chmod +x it and put it under /etc/init dir. Upstart information

not sure if this is a silly question...apples and oranges maybe..but...

For Windows I used to download either 32 bit binaries from ASF site or 64 bit binaries from Apache lounge site (for no particular reason). So - that was the way I new what version I have.

I've switched to ubuntu (for educational purposes) and got used to compiling from source...well...you know...make..make install etc.

mmmm

1) So - when I compile apache server from source - what binaries do i get? 32? 64?

2) Are there any benefits from having 64 bit Apache binaries? any cons?

Thnaks guys:)