Super slow backup restoration

V

ViAdCk

Verified User
Joined
Feb 14, 2005
Messages
250
Hello,

On a couple of new installed servers with directadmin we are seeing super slow backup restoration. In one case it is taking almost 30 minutes to restore 5 accounts with a total size of 500MB.

For example it halts during a couple of minutes with every user with the following message: Ensuring stored domains are present

It also stops for a couple of minutes with the message "Restoring Databases" but nothing is happening when I check the server on the console.

This isn't a hardware issue as this has been tested on various different servers.

Any idea what could be going on?

Regards
 
I have reverted back to traditional tar.gz backup files but the issue persists.

It hangs during a couple of minutes on "Ensuring stored domains are present" and then it hangs a couple of minutes on "Restoring Databases" (even if the backup doesn't have any databases), then it hangs on "Extracting ~/imap directory".

This happens on every restoration, even if the backup is only 5MB in size. A small user that normally restores in 30 seconds now takes 5 minutes.
 
Last edited:
same here, restore of a 60gb backup on SAS SSD Raid 10 took 4 hours (creation on cPanel box just 18mins, on SAS 10k drives)
 
what you see from ssh in TOP, ATOP during restoration? what processes with which parameters? disk load? idle? sy? us? etc..
 
Zhenyapan said:
what you see from ssh in TOP, ATOP during restoration? what processes with which parameters? disk load? idle? sy? us? etc..
Click to expand...
As I said in my original post, nothing happens in the console when it hangs. The load goes to 0.00 with no processes at all.
 
you start restoration, see load increase and extracting in TOP, process gone, you go to /home/restoreduser and see restored files, but there still no message from DA? maybe problem not with archive but with DA messaging or dataskq or quota recheck? try to unpack this archive from console tar xzf archivename.tar.gz
 
Zhenyapan said:
you start restoration, see load increase and extracting in TOP, process gone, you go to /home/restoreduser and see restored files, but there still no message from DA? maybe problem not with archive but with DA messaging or dataskq or quota recheck? try to unpack this archive from console tar xzf archivename.tar.gz
Click to expand...
I have already said what happens:

It hangs during a couple of minutes on "Ensuring stored domains are present" and then it hangs a couple of minutes on "Restoring Databases" (even if the backup doesn't have any databases), then it hangs on "Extracting ~/imap directory".

When it hangs, nothing happens on the console. This happens with every backup, doesn't matter if it's 5MB or 2GB. This is the reason why a 5MB backup takes around 5 to be restored completely.

This only happens when restoring the backup, backup generation works as usual.
 

Attachments

  • Screenshot 2021-10-16 at 12.34.15.png
    Screenshot 2021-10-16 at 12.34.15.png
    37.9 KB · Views: 139
strange, enough memory? maybe it uses some /tmp partition? maybe CSF process tracking kills it?
unpack this archive from console tar xzf archivename.tar.gz works without errors?
 
Zhenyapan said:
strange, enough memory? maybe it uses some /tmp partition? maybe CSF process tracking kills it?
unpack this archive from console tar xzf archivename.tar.gz works without errors?
Click to expand...
Unpacking the backup from the command line happens in seconds.

CSF is disabled and there is plenty of ram/cpu available. When it halts resource consumption goes to 0.
 
I have the same problem with restoring backups even with only databases+ database settings (over 360 domain configuration files), only 4.5gb space (after unpack) and took nearly 2 hours....even on server with 48 cores/96 threads, nvme disk and 768 gb ram (here maybe over 1 hour)....
Any ideas?
@ViAdCk did you resolve that?
@smtalk maybe you can help?
 
Last edited:
ShinJii said:
I have the same problem with restoring backups even with only databases+ database settings (over 360 domain configuration files), only 4.5gb space (after unpack) and took nearly 2 hours....even on server with 48 cores/96 threads, nvme disk and 768 gb ram (here maybe over 1 hour)....
Any ideas?
@ViAdCk did you resolve that?
@smtalk maybe you can help?
Click to expand...
tar.gz? when you restore it manually from ssh - it fast? mysql import from ssh fast too? try install pigz via custombuild and set it to 4 cores in directadmin.conf
 
You must log in or register to reply here.
Back
Top