Restoring Horde preferences.... - stuck many minutes

[postcd]

Is it normal that when i restore fullbackup file from "Home »Backup »Restore a Full Backup/cpmove File"

the process delays many minutes on message "Restoring Horde preferences...." ?
why is it happening, any way to speedup this?

 

[Bashed]

Same issue here

 

[cPanelMichael]

Hello

Does it end up restoring successfully? What are the server load and iowait values at the time of the restore?

Thank you.

 

[postcd]

Does it end up restoring successfully? What are the server load and iowait values at the time of the restore?

Yes, it always succeed, but even 11MB tar.gz fullbackup file takes around 20 minutes stuck at this message

top - 12:30:38 up 1 day, 6:59, 1 user, load average: 6.03, 5.38, 4.52
Tasks: 71 total, 2 running, 68 sleeping, 0 stopped, 1 zombie
Cpu(s): 31.1%us, 3.8%sy, 0.0%ni, 33.1%id, 31.9%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 13312000k total, 1581724k used, 11730276k free, 0k buffers
Swap: 0k total, 0k used, 0k free, 0k cached

I was already preventivelly set Critical load threshold in WHM Tweak settings to 40, but dont help..
Configserverfirewall disable dont help

This is from cpanel error log:

[2014-04-30 12:35:18 -0400] warn [servers_queue] Cannot queue task dbstoregrants miccjgxs ( no id, probably a duplicate ) at /usr/local/cpanel/Cpanel/LoggerAdapter.pm line 26
Cpanel::LoggerAdapter::warn(Cpanel::LoggerAdapter=HASH(0x200e7748), 'Cannot queue task dbstoregrants miccjgxs ( no id, probably a duplicate )') called at /usr/local/cpanel/bin/servers_queue line 90
main::queue_tasks('dbstoregrants miccjgxs') called at /usr/local/cpanel/bin/servers_queue line 75
cphulkd rejected registration attempt for whostmgrd with: 500 AUTH FAILED

seems happening for each restored account

 

[cPanelMichael]

We do have an internal case open (92317) to investigate those warning messages during account restores. However, you mentioned the restore completed successfully, and the only issue was with the time it took for the restore. The load average you provided seems a bit high. Do you notice any change in the time it takes to restore the account if the load is lower?

Thank you.

 

[postcd]

> Do you notice any change in the time it takes to restore the account if the load is lower?

today im restoring around 15 accounts and some stuck for 6 minutes and some (a big one) around 20 minutes. Its strange. Thats why im asking if i can restore paralelly. That big one is like only 150mb .tar.gz

 

[cPanelMichael]

today im restoring around 15 accounts and some stuck for 6 minutes and some (a big one) around 20 minutes. Its strange.

To clarify, was the server load already at an elevated state when you started the restore? I could not reproduce this problem on a test machine, but you are welcome to submit a ticket so we can take a closer look. Note that your other question is answered on the thread you referenced.

Thank you.

 

[Bashed]

Still persists (minimal load, less than 1 CPU / 30% memory).

Cpanel 11.42.1 build 13

 

[cPanelMichael]

Feel free to open a support ticket so we can take a closer look. You can post the ticket number here so we can update this thread with the outcome.

Thank you.

 

[Jcats]

We are experiencing this as well. Load under 1, quad core E3 server, plenty of free memory, raid10 disks no IO. If you check out the mysql processes during the backup you will notice the issue is with the temp database that looks to be created. I notice most of the queries containing mnemo which is hordes notes and memos applications.

| 16661811 | horde | localhost | cptmpdb_horde_G34JG5WK58doYu4F | Sleep | 132 | | |
| 16661816 | horde | localhost:49215 | cptmpdb_horde_G34JG5WK58doYu4F | Sleep | 132 | | |
| 16661817 | horde | localhost:49216 | cptmpdb_horde_G34JG5WK58doYu4F | Query | 4 | committing alter table to storage engine | ALTER TABLE `mnemo_sharesng` ADD `share_parents` text |
| 16661953 | horde | localhost:49313 | cptmpdb_horde_G34JG5WK58doYu4F | Sleep | 65 | | |

- - - Updated - - -

To note, it is causing a considerable amount of IO on the server during that time

- - - Updated - - -

I take that back about mnemo being mostly noticed, its just the horde queries in general, just did another restore and horde took 170 seconds, start to finish.

| 86506 | horde | localhost | cptmpdb_horde_Ci8yp2E2bGMzOhO0 | Sleep | 164 | | |
| 97852 | horde | localhost:59644 | cptmpdb_horde_Ci8yp2E2bGMzOhO0 | Sleep | 163 | | |
| 97853 | horde | localhost:59645 | cptmpdb_horde_Ci8yp2E2bGMzOhO0 | Query | 0 | creating table | ALTER TABLE `turba_objects` ADD `object_manager` varchar(255) |
| 97935 | horde | localhost:59651 | cptmpdb_horde_Ci8yp2E2bGMzOhO0 | Sleep | 115 | | |

 

[WhiteDog]

Having the same issue on all our servers. For example, moving an account under 1MB takes about 4 minutes now.
This is to a new and empty server (Quad-Core, RAID10, lots of memory). The issue is definitly not with server load.

This is extremely annoying, should we need to restore 200+ accounts to a new server for whatever reason it will add half a day (!) to the restore time.

EDIT: I also tested with Horde disabled on the new server.

You then get an error on restore:
DBD::mysql::st execute failed: Table 'cptmpdb_horde_JPvsVoNLbpevoSCo.horde_dav_collections' doesn't exist at /usr/local/cpanel/Cpanel/Horde/Utils.pm line 77. DBD::mysql::st execute failed: Table 'cptmpdb_horde_JPvsVoNLbpevoSCo.horde_dav_collections' doesn't exist at /usr/local/cpanel/Cpanel/Horde/Utils.pm line 77. Failed to restore horde:

But the Horde part only takes a few seconds. If you need to restore many accounts, I would suggest disabeling Horde during the restore as doing speedy restores will probably outweigh losing a few settings at that time...

 

[scottc]

Same problem here. It can take 10 or 15 minutes to restore a small cpanel account.

 

[cPanelMichael]

I've been unable to find any support tickets open for this issue. Please ensure you open a support ticket if you are experiencing this issue so we can take a closer look and investigate further. Post the ticket number here so we can update this thread with the outcome.

Thank you.

 

[Jcats]

#4896925

One of the techs is trying to reproduce it, but having no luck. Maybe we are all running something similar, but this seems completely mysql related, we are running:

cloudlinux(not running dbgovernor)
litespeed
mysql 5.6

 

[konrath]

Same here, after cpanel update to version WHM 11.42.1 (build 12)

I am migrating an old server accounts with WHM version 11.38.2 (build 12)

The problem must be happening due to the conversion of old WHM accounts to a new WHM accounts?

Thank you
Konrath

 

[cPanelPeter]

Hello,

Just to update, on ticket 4896925 it was determined that the source server had an older version of Horde installed then the destination server. This means that the users are having to have their databases converted to the newer format, this would likely need to be done for each user regardless of account size.

Can anyone else confirm the version of Horde on both servers?

 

[konrath]

Horde 3.3.13 ( old server )

Imp H5 (6.1.4) ( new server )

That's what I thought.

Thank you
Konrath

 

[cPanelMichael]

An increase in restore time for the conversion to the new Horde version is to be expected. You should not experience the same issue if you restoring an account that was packaged on a server that already uses the new version of Horde included with cPanel.

Thank you.

 

[konrath]

I need 3000 accounts migrate from an old server. It is impossible with to horde enabled ..

1) I did an experiment. Deactivated the horde on the new server.
2) I quickly restored an account from the old server to the new server.
3) I activated the horde on the new server.
4) The horde webmail work without problems. The emails in this account test worked well in horde.

If you need to migrate thousands of accounts of old servers (server crash for example), turn off the horde and up again after restoring the accounts.

Thank you
Konrath

 

[BianchiDude]

Same issue here, is there still an open case for this?

Pretty difficult to restore a server with a large number of account having to wait 20+ minutes each time for it to restore horde.