admin backup with httpd shut down

B

bernhard

Verified User
Joined
Nov 19, 2012
Messages
31
Location
Oldebroek, The Netherlands
I have a quick question. Is it possible to start an admin backup with httpd shut down? I want to reinstall one of my servers because it's giving me problems that I have been told can only be repaired with a reinstall. I want to make an admin backup but want to be sure that after I have done the backup no new data is placed on the server that would be lost.
 
Hi, that shouldn't be a problem.
Go into DA -> click services -> Click to stop Apache or Nginx, make sure the processes have actually stopped.
Then go into the Admin Backup and add the backup to the queue.
 
Please explain and document We've been running CentOS since 3.x, and are now on latest CentOS6, and we've never seen this behavior. (On reboot, yes, but not in the normal course of events.)

Jeff
 
What do you want documented Jeff? The thing about Exim?
I did not test anymore on Cenos 6, but I also disabled Exim via the manager in DA and indeed after a little while it was running again without me starting it and without rebooting the server.
I have no clue why, but like I said on Centos 5 I experienced the same behaviour (have not tested on 6).
 
You should also just suspend the users for be sure no-new-data are insert on DB/FTP/Email.

Maybe using a custom Suspend page that explain what is going on to the users who visit websites.

Regards
 
I had a time a suspende account was not backupped, in spite of the fact that the option to not backup suspende accounts, was not selected.
So normally in these occasions, all our users get an email we are backupping the server for transfer to another server, with the request to not change anything anymore after a certain time, mostly at night.
Then I stop Exim and Mysql and start the backup and transfer everything to the new server. This works very good. And since I'm awake most of the time at night, it's no problem for me.
 
I am still concerned as to why/how exim would restart. It shouldn't. I've never seen it restart. I have a DirectAdmin powred CentOS6 system running but not hosting any accounts. I stopped exim on it a few weeks ago and it has not restarted.

Perhaps some function (adding/removing a user/domain/account) is causing it to restart when it shouldn't.

As far as moving accounts without losing data:

I don't like suspending accounts on the o9ld server to move them to a new server; it can cause a lot of downtime if the move includes a lot of sites. I've also occasionally had backup/restore problems on suspended accounts. But some sites will need to be suspended. If so, I suspend the entire user, mainly because individual site suspension in DirectAdmin is relatively new and I'm not sure if all bugs have been worked out of it yet.

We let users know before moving accounts, as far in advance as possible. We give them a time-frame during which they should NOT attempt to change static account data.

Identify (as your users) all accounts which have visitor-modiable content (blogs, forums, etc.), and those only you'll probably need to suspend or have users either suspend the site or it's modirfication features during the account move.

Then in this order, at the appointed time/date:

1) change DNS TTL on all zones to 600 (ten minutes)

1) move accounts, but not point DNS to new server as of yet.

2) suspend sites identified above as needing suspension, on both servers.

3) move DNS to new server.

4) Move (again) but only databases and email to new server.

5) Unsuspend sites on new server as necessary

5) Six to 12 hours later again move email only to new server to catch any unmoved email delivered to the old server because of sending server queue/cache issues.

Note this is a work in progress, defined over the past few days because we're moving all our clients to a new hosting latform over the next six weeks. It may need some refining, and your input is welcome.

Thanks.

Jeff
 
I think I remember why my Exim restarted. I just stopped it by using "/etc/rc.d/init.d/exim stop" which indeed stopped Exim. But I did not use the stop option in DA, maybe that cause DA to start Exim again a bit later.
Unfortunately I don't remember if I also tryed stopping Exim from DA, I think I did, but I'm not sure anymore.

I don't have any more input because we don't move that way. We just stop the mailserver. And because mail should be offered several times before getting deleted by an MTA, all should be well when the DNS is on the next server.
Next to that our users can choose to have a backup-mx on their domain, so mail won't get lost when the server would be out and down.
Not every customer chooses for that option (which only costs 1 or 2 euro a year more), so I don't mind a lot if 1 or 2 customer emails get lost (which never happened until now). And all business accounts do have taken a backup-mx.
 
Richard G said:
And because mail should be offered several times before getting deleted by an MTA, all should be well when the DNS is on the next server.
Click to expand...
All should be well. A few year ago several MTAs wouldn't check DNS again, but retry only on the same DNS. I don't know if that's been fixed or not. If I recall correctly, one of the problem servers was qmail, but I could be wrong.

Jeff
 
but retry only on the same DNS.
Click to expand...
That would not be a problem because when you move servers and use the same DNS name of the old server which we always do (ns1.domain.com etc.), the mail would then be delivered to the new server, because only the ip changes.
Unless you mean that the retry would only be done to the same ip.

If yes, then there could be a problem on very few occasions, but then again. I don't care. If customers are not prepared to pay 1 euro per year for a backup-mx, they should not complaint.
 
Last edited:
I should have written but retry only on the same IP# because that is indeed the problem. When we move sites between servers we always move email again about 12 hours later, just to make sure we catch straggling email.

Note I'm currently moving a lot of clients from an old infrastructure to a new one, so this is an ongoing project with ongoing tweaking, on a day-by-day basis. Ask me again in a few months :).

Jeff

Jeff
 
You must log in or register to reply here.
Back
Top