Editorial: TAKE TIME TO BACKUP

[jholder]

This is a little editorial:

It is good admin etiquette to backup. In the open source edition, it doesn't include and automated backup process, but you CAN backup your files!

All information is most likely located in the /opt/zimbra folder. If you stop zimbra, and simply tar up the folder stick it in a different location, then you are good to go!

I cannot stress enough to our valued users, TAKE TIME TO BACKUP!!!

Let's take a failed upgrade as an example.

If you upgrade from 4.0.x to 4.5 and it fails (for any reason), it may be possible to install your 4.0.x over your failed upgrade and get it working. HOWEVER, you now *may* have tables in the db that are not active, and may cause you future upgrade headaches.

We love helping, and providing the very best support in the open source community, but it's important that you keep your info safe.

So, to sum it up: before you perform any sysadmin maintenance like upgrades, injections, or anything else, take time to backup.

You may be down a little longer, but you'll be better in the end.

Sincerely,
john

[Klug]

Reading this Zimbra blog article is a nice idea too : http://www.zimbra.com/blog/archives/..._b_1.html#more

[yang]

duh, if you don't do backups before you do upgrades, you are probably not the right person to be doing the upgrade ... time to find a half way decent SA

[cvidal]

For those using the OS edition, here is a simple backup script that performs very well.

It stops Zimbra, does an rsync to a backup directory, and starts zimbra again. The backup directory can then be put on your regular backup system.

Simply copy the script to /etc/cron.daily so it gets executed every night.

For systems with up to 500 users the total execution time is of a maximum of 10 minutes, what means that your system will be out of service 10 minutes a day, a reasonable price for the OS edition (IMHO).

Carlos

[phoenix]

Hi Carlos

Did you forget the script/attachment?

[Klug]

Originally Posted by cvidal

For systems with up to 500 users the total execution time is of a maximum of 10 minutes, what means that your system will be out of service 10 minutes a day, a reasonable price for the OS edition (IMHO).

As long as you have two decent servers, a decent network and the script

[tyco]

where is that script?

thanx

[JoshuaPrismon]

Originally Posted by jholder

This is a little editorial:

It is good admin etiquette to backup. In the open source edition, it doesn't include and automated backup process, but you CAN backup your files!

All information is most likely located in the /opt/zimbra folder. If you stop zimbra, and simply tar up the folder stick it in a different location, then you are good to go!

I cannot stress enough to our valued users, TAKE TIME TO BACKUP!!!

Let's take a failed upgrade as an example.

If you upgrade from 4.0.x to 4.5 and it fails (for any reason), it may be possible to install your 4.0.x over your failed upgrade and get it working. HOWEVER, you now *may* have tables in the db that are not active, and may cause you future upgrade headaches.

We love helping, and providing the very best support in the open source community, but it's important that you keep your info safe.

So, to sum it up: before you perform any sysadmin maintenance like upgrades, injections, or anything else, take time to backup.

You may be down a little longer, but you'll be better in the end.

Sincerely,
john

This is the one area that I think the Zimbra folks have it unequivocally wrong. The difficulty in backing up Zimbra stems from the fact that this feature set was restricted to the network edition only.

I don't know of any large scale (100+) mailbox Open Source installations. I think everyone corporate is pretty much mandated to go to NE if they have a large corporate base. I really feel that long term the lack of the live and per mailbox backups is a huge issue for the open source Zimbra community. I think witholding this feature makes it less likely that there will be robust backups and more likely that there will be a catastrophic failure.

Finally LVM snapshotting is your friend.

Edit: To be fair to Zimbra folks, this is scoped more closely to upgrading. cp -r is a definite must before doing any sort of upgrade, NE or OS. I know that mailbox by mailbox backup/restore is typically a enterprise activity, but I think there needs to be more thought behind mandating a standard way to backup Zimbra OS edition. Even if it is just "always make /opt a lvm with snapshot ability. Back in my IT consulting days, I can't count the number of times that we had people who thought they had a good backup, but never tested them, or just happened to have a failed backup right before a major outage.

[jholder]

Keep in mind that we are a company with bills to pay.
For example, we pay to host these forums, and pay to develop Zimbra.

We can't pay those bills if we give away everything.

[JoshuaPrismon]

Originally Posted by jholder

Keep in mind that we are a company with bills to pay.
For example, we pay to host these forums, and pay to develop Zimbra.

We can't pay those bills if we give away everything.

No one is asking you too. I just suspect that product support in general, clustering, etc might be areas that really have a lot of value to paying customers, while the lack of coherent backups may turn out to be a huge detriment to the open source platform.