Koozali.org: home of the SME Server
Obsolete Releases => SME Server 8.x => Topic started by: abarron on April 30, 2015, 02:13:03 AM
-
Searched forum but have not found exactly this issue.
First I tried to upgrade my server from v8.1 to V9.0 and for some reason I couldn't make this work. So I did a fresh install of v8.0 and did all the updates to get v8.1.
Then I started a restore from workstation (which is taking an inordinately long time incidentally) and found that I could no longer access server manager from my browser. I am able to log into root via a terminal session using PuTTY.
I am not very familiar with Linux and I am trying to tell if the restore is finished by using Wireshark and monitoring the traffic between the server and the workstation (running it on the workstation).
Can I tell if the restore is finished in a better way? Is there something I can look at via my tty session?
Also any ideas why I can't access server manager.
Thanks in advance for any help.
-
abarron
Try accessing server manager directly at the server console.
Log in as admin (at the command prompt) & from the menu select server manager.
It will open a text based server manager view
-
Thank you Janet. Yes I can access server manager from the console via my terminal connection.
That is I can access the login dialogue and when I supply the appropriate credentials it brings up a new URL and then eventually cycles back to the login dialogue after a few seconds.
This is essentially the same behaviour I am getting from the web interface.
Any more ideas gratefully received.
-
are you still waiting for the data restore?
-
Yes Stefano I am, lot of data.
-
abarron
Yes I can access server manager from the console via my terminal connection.
That is I can access the login dialogue and when I supply the appropriate credentials it brings up a new URL and then eventually cycles back to the login dialogue after a few seconds.
This is essentially the same behaviour I am getting from the web interface.
I do not call that being able to access server manager. I call that NOT being able to access server manager.
Did you configure the identical backup job in server manager, so it can find your backup to do the restore from ?
-
how much data?
is your raid rebuilding? raid1? raid5?
how is connected the remote machine? 100 MBit? 1Gbit?
you don't need server-manager to monitor restoring..
just login as root and use top, iptraf, df (open at least 3 different sessions)
you'll see what's running (dar), network traffic speed (iptraf), disk space occupation (df)
in the meanwhile, I suggest you to smoke a lot of cigarettes and drink coffee :-)
if you are not a smoker, just start ;-)
-
abarron
I do not call that being able to access server manager. I call that NOT being able to access server manager.
Did you configure the identical backup job in server manager, so it can find your backup to do the restore from ?
Yes you are right, sorry.
Yes I had configured the backup the same way, it offered me a selection of backups to restore starting with the last full
backup which is what is currently restoring.
Thank you for your help
-
how much data?
is your raid rebuilding? raid1? raid5?
how is connected the remote machine? 100 MBit? 1Gbit?
you don't need server-manager to monitor restoring..
just login as root and use top, iptraf, df (open at least 3 different sessions)
you'll see what's running (dar), network traffic speed (iptraf), disk space occupation (df)
in the meanwhile, I suggest you to smoke a lot of cigarettes and drink coffee :-)
if you are not a smoker, just start ;-)
About 750 GB of data
Top shows Dar is running and also MD2_Raid5 is that a raid rebuild?
100Mb network
IPTraf shows traffic that pretty much agrees with traffic from workstation shown by Wireshark.
Guess I have to take up smoking and wait :shock:
Thank you for your help.
-
well
750 Gb -> about 75000 MB
100Mb -> about 10 MB/s
then you need at least 75000/10 s -> 22 hours
don't hold your breath ;-)
-
Ah! Currently it has been running for about 70 hours.
-
well, I forgot to say that reid5 rebuilding is a big stress..
please, post the output of
df -h
and the first 5 rows of
top
-
abarron
Yes that's a lot of data to restore over a 100Mb network & not really practical to do so.
Currently it has been running for about 70 hours.
It may have hung & you are waiting for nothing. Use the tools Stefano advised to check what is happeneing.
You would be far better off (IMHO) to copy your backup off the workstation to a USB drive (locally connected on sme server formatted as ext3/4) to a correct folder location to suit the backup "layout", & then use that USB drive to restore from on the sme server.
That amount of data might typically take 8 - 10 hours to restore depending on various factors (USB speed, data type etc).
-
abarron
...I tried to upgrade my server from v8.1 to V9.0 and for some reason I couldn't make this work
sme 9 (CentOS 6.5) does not support direct upgrades from prior versions, so the only path to update an older version of sme server, is via backup & restore.
Can I tell if the restore is finished in a better way?
Server manager restore panel should display the prgress of the restore & tell you when it is complete, I assume you instigated the restore via server manager & then server manager disconnected or you closed it (???).
-
abarron
top shows .... MD2_Raid5 is that a raid rebuild?
Also show output of
cat /proc/mdstat
If your restore is nearing completion, then you may be best to just wait, assuming it is still running.
-
Stefano,
Top output.
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
19383 root 15 0 4864 720 488 S 24.7 0.1 5:00.38 iptraf
3181 root 18 0 734m 354m 832 R 6.5 70.6 1332:27 dar
321 root 10 -5 0 0 0 S 2.6 0.0 139:42.24 md2_raid5
3145 root 10 -5 0 0 0 S 2.0 0.0 35:14.62 cifsd
19437 root 15 0 2440 1064 800 R 2.0 0.2 0:00.26 top
Output from DF -h
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/main-root
451G 127G 301G 30% /
/dev/md1 99M 19M 76M 20% /boot
tmpfs 252M 0 252M 0% /dev/shm
192.168.7.4:smebackup
932G 196G 737G 21% /mnt/smb
Cheers.
-
post also the top headers values (load etc)
I that you've copied only 126 GB of data...
take a look at /var/log/messages too
-
abarron
sme 9 (CentOS 6.5) does not support direct upgrades from prior versions, so the only path to update an older version of sme server, is via backup & restore.
That is how I was going to try and do it, I had read that I had to do a fresh install.
Server manager restore panel should display the prgress of the restore & tell you when it is complete, I assume you instigated the restore via server manager & then server manager disconnected or you closed it (???).
Yes that is how I started it. That is also why I am at a loss as to know when the restore has finished as I am unable to access server manager
-
post also the top headers values (load etc)
I that you've copied only 126 GB of data...
take a look at /var/log/messages too
top - 20:34:31 up 2 days, 20:33, 3 users, load average: 1.75, 1.92, 1.91
Tasks: 174 total, 3 running, 171 sleeping, 0 stopped, 0 zombie
Cpu(s): 11.1%us, 18.6%sy, 0.0%ni, 3.3%id, 65.1%wa, 1.6%hi, 0.3%si, 0.0%st
Mem: 514860k total, 503276k used, 11584k free, 236k buffers
Swap: 1048568k total, 735404k used, 313164k free, 12512k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3181 root 18 0 734m 422m 1116 R 15.8 84.1 1334:41 dar
321 root 10 -5 0 0 0 S 4.8 0.0 140:26.83 md2_raid5
19383 root 15 0 4864 720 488 S 2.3 0.1 5:31.44 iptraf
19355 root 15 0 10180 624 512 S 1.9 0.1 1:10.56 sshd
19445 root 16 0 2440 696 456 R 1.9 0.1 0:02.73 top
I also notice that the headers seem to say that system has been up for only two days. I started the restore on 27th. Or does it say 2 days and 20:33 hours?
-
ok.. let top run and take a look at wa (waiting) value..
I see it's 65%
that value should be as low as possibile..
-
abarron
Also show output of
cat /proc/mdstat
If your restore is nearing completion, then you may be best to just wait, assuming it is still running.
Sorry Janet I missed this request, here it is.
cat /proc/mdstat
Personalities : [raid6] [raid5] [raid4] [raid1]
md1 : active raid1 hda1[0] hdb1[1] hde1[2] hdf1[3]
104320 blocks [4/4] [UUUU]
md2 : active raid5 hdf2[2] hde2[3](S) hdb2[1] hda2[0]
488182784 blocks level 5, 256k chunk, algorithm 2 [3/3] [UUU]
unused devices: <none>
Thank you for your help it is much appreciated.
-
abarron
Yes that's a lot of data to restore over a 100Mb network & not really practical to do so.
It may have hung & you are waiting for nothing. Use the tools Stefano advised to check what is happeneing.
You would be far better off (IMHO) to copy your backup off the workstation to a USB drive (locally connected on sme server formatted as ext3/4) to a correct folder location to suit the backup "layout", & then use that USB drive to restore from on the sme server.
That amount of data might typically take 8 - 10 hours to restore depending on various factors (USB speed, data type etc).
If I can find out that the restore has stopped I will do just that. I do remember trying to attach a USB drive to the server before and not being able to get it to work, but I will try again.
What do you mean by: to a correct folder location to suit the backup "layout"
Thanks.
-
abarron
What do you mean by: to a correct folder location to suit the backup "layout"
When a backup is run say to a local USB drive (or workstation) it creates a certain folder structure, something like the server domain name, so if you manually move the backup file (only), then you need to create the same folder structure on the USB drive, so that the backup/restore function will correctly find the backup file.
The folder structure should be the samne as shown on your wporkstation backup share.
USB drives need to be formatted to ext3 or ext4, refer the Howto section of this site for details.
-
abarron
When a backup is run say to a local USB drive (or workstation) it creates a certain folder structure,
<snip>
USB drives need to be formatted to ext3 or ext4, refer the Howto section of this site for details.
Oh I understand, thank you.
So when I get the USB drive working I guess I then have to mount the work station drive, that contains the backup, to the server manually. Then I have to copy the backup files to the USB drive over the local LAN.
The restore does still seem to be running, but do you think I would be better off doing it the local USB route anyway?
TY again.
-
abarron
Stefano suggest you have only copied 126Gb, so you stil have a long way to go.
That has taken 3 days already so do you want to wait another 6 days or so ?
If something fails ie network connection, then all is lost.
You can reinstall the sme OS in 15 minutes, configure the backup in 5 minutes, transfer backup file in say an hour, then, start the backup which might finished in half a day or so, much quicker than waiting another 6 days in my books.
Your choice though.
It also seems to me that you do not have enough space on your sme server to restore that data
ie
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/main-root
451G 127G 301G 30% /
/dev/md1 99M 19M 76M 20% /boot
tmpfs 252M 0 252M 0% /dev/shm
192.168.7.4:smebackup
932G 196G 737G 21% /mnt/smb
& you are trying to restore 737Gb (which is compressed) so more than that probably.
I am not sure if the actual size of the RAID 5 array is being shown.
-
Ok Janet I am going to take your advice and try and set up a USB disk.
I just realised I am guilty of a typo, the uncompressed data size is actually only about 350GB from memory. Compressed should have been 250GB not 750GB. That makes a significant difference but still only about half way through so I am going for USB.
If I can get that to work I will do my backups that way from now on.
Thank you for all your help, I really appreciate it.