Ubuntu – Should use EXT4 or XFS to be able to ‘sync’/backup to S3

amazon s3ext4Ubuntuweb-serverxfs

It's my first message here, so bear with me…

(I have already checked quite a few of the "Related Questions" suggested by the editor)

Here's the setup,

a brand new dedicated server (8GB RAM, some 140+ GB disk, Raid 1 via HW controller, 15000 RPM)
it's a production web server (with MySQL in it, too, not just serving web requests); not a personal desktop computer or similar.
Ubuntu Server 64bit 10.04 LTS

We have an Amazon EC2+EBS setup with the EBS volume formatted as XFS for easily taking snapshots to S3, via AWS' console.

We are now migrating to the dedicated server and I want to be able to backup our data to Amazon's S3. The main reason being the possibility of using the latest snapshot from an EC2 instance in case of hardware failure on the dedicated server.

There are two approaches I am thinking of:

do a "simple" file-based backup with rsync, dumping the database' and other files, and uploading to amazon via S3 API commands, or to an EC2 instance, or something.
do a file-system "freeze" (using XFS) with the usual ebs/ec2 snapshot tool to take part of the file system, take a snapshot, and upload it to Amazon.

Here's my question (or series of questions):

Can I safely use XFS for the whole system as the main and only format on the dedicated server?
If not, is it safe to use EXT4? Or should I use something else?
would then be possible to make snapshots of the system to upload to Amazon?
Is it possible/feasible/practical to do what I want to do, anyway?
any recommendations?

When searching around for S3/EBS/XFS, anything relevant to my problem is usually focused on taking snapshots of a XFS system that is already an EBS volume. My intention is to do it in a "real"/metal dedicated server.

Update: I just saw this on Wikipedia:

XFS does not provide direct support for snapshots, as it expects the snapshot process to be implemented by the volume manager.

I had always assumed that I could choose 2 ways of doing snapshots: via LVM or via XFS (without LVM). After reading this, I realize these 2 options are more like it:

With XFS: 1) do xfs_freeze; 2) copy the frozen files via, eg, rsync; 3) unfreeze xfs
With LVM and XFS: 1) do xfs_freeze; 2) make a binary copy of the frozen fs via lvcreate and related commands; 3) unfreeze xfs; 4) somehow backup the LVM snapshot.

Thanks a lot in advance,

Let me know if I need to clarify something.

Best Answer

Every linux filesystem (ext2, ext3, ext4, xfs, jfs, raiserfs) in current kernel can be frozen, but it must be placed on LVM first.

If you have LVM, making a snapshot automatically freezes the FS for the time it takes to make the snapshot -- it's better than doing only a freeze (your data is still available for writing and will not break the backup) and much better than simple rsync (as it will copy files in a consistent state).

Related Solutions

Cron – Ubuntu 9.10 cron script ‘ec2-consistent-snapshot’ unable to execute and access files

sudo crontab -e edits root's crontab right? When you say you have $HOME/.awssecret, what is $HOME? root's home or yours?

You might want to think about using /etc/cron.d, you can additionally add the name of the user to execute the script as in these files (e.g. caution: slightly different syntax for the scripts)

Edit (answering your question in the comment):

Create a file /etc/cron.d/myEc2Crontab

make it contain:

SHELL=/bin/sh
PATH=whatever you need as your path
12 18 4 2 *  root ec2-consistent-snapshot --mysql --xfs-filesystem /vol vol-xxxxxxx >>/mnt/backup.log 2>&1

note the added 'root' just after the time specification, prior to your command. This specifies the user the command runs as.

Linux – Converting Amazon EC2 ext4 EBS volume into XFS filesystem

You can either convert your root volume to XFS or create a second volume. Personally, I would favour the latter as it offers more flexibility.

Option 1: Convert Root Volume to XFS

Stop your instance (not terminate) (let's call it instance A)
Start a new instance (let's call it instance B)
Detach the root volume from instance A and attach it to instance B
Create a second EBS volume, attach it to instance B
Format the new EBS volume as XFS (install xfsprogs if not already done)
Copy all the data from the first volume to the new one (e.g. using rsync -aHAXxSP /source /target)
Detach the new volume from instance B and attach it as the root volume of instance A
Start instance A
Terminate instance B (the original root volume should persist, keep it around until things work to your liking).

The reason for attaching the root volume to another instance is to attain consistency, which would be difficult with the volume in use.

Option 2: Move data to a second EBS volume

Create a second EBS volume and attach it to your instance; format it as XFS and mount it
Identify which directories you wish to move to the new volume (some to consider include: /var/log, /var/lib/mysql, /var/www, /var/spool/mail, /var/vmail)
Stop as many services as possible to remove write locks
Use lsof | grep /path/to/dir to check for remaining write locks
Move the directory to the new EBS volume
Mount bind the new directory to its old location (i.e. mount -o bind /mnt/path/to/dir /orig/path/to/dir)
Repeat for each directory
Start your services to ensure all is working
Edit your /etc/fstab file to make the mount points permanent; e.g.: /mnt/path/to/dir /orig/path/to/dir bind defaults,noatime,bind 0 0

Restart to ensure everything persists and functions as it should. You may also want to consider shrinking your root volume down a bit since, hopefully, it shouldn't be growing (or changing) much with all the data removed from it.

(As a point of mention, it may be possible to freeze an ext4 file system using fsfreeze, which is included in util-linux-ng).

Best Answer

Related Solutions

Cron – Ubuntu 9.10 cron script ‘ec2-consistent-snapshot’ unable to execute and access files

Linux – Converting Amazon EC2 ext4 EBS volume into XFS filesystem

Related Topic