ZFS Replication - Systems Maintenance(Archived)

shall i possible to mount remote server through replicating snapshot, whether snapshot only or data also replicated in remote server? give some commands 

Cross-posted:
[http://forums.sun.com/thread.jspa?threadID=5436050|http://forums.sun.com/thread.jspa?threadID=5436050] 

Thank you rukbat ...

Related

changing from mysql cluster to remote mysql

Hi anybody,
I'm running a high availability cluster (3 esx guests) with embedded mysql cluster, which turned out is quite unstable and also not supported (running mysql cluster on ESX). Anybody knows how to migrate to Remote Mysql with keeping the current SRS Server settings (utpreserve ?) and also the current mysql db entries.
btw, I'm running vdi 3.2.2, the latest one.
kind regards, thomas
PS: Much more interesting for me would be a migration to a Oracle schema, but this is another topic not existing so far. 
I would assume you would be able to backup up the current database, reconfigure VDI and then restore the database. I haven't tried this nor is there a wiki page on this particular reconfiguration so my assumption could be wrong. 
the problem is that there is no reconfigure. i need do to a vda-install - u which wipe away my srs settings.
regards, thomas 
Depending on what you have in your SRSS datastore you might be able to accomplish it by backup the VDI DB "vda backup", then backup the tokens and desktops using "utuser -o >> file" and "utdesktop -o >> file". Then unconfigure VDI, vda-config -u and then reconfigure, "vda-config". Restore the VDI DB "vda-restore 'backup/file'" Then restore the tokens and desktops "utuser -af file" and "utdesktop -af file".
If you have more custom things in the datastore than tokens and desktops the above may not work. I don't know if a the vda-restore command works with backups from local DBs to remote DBs either. 
Hi Zettabyte,
so I have gone through all the pain of migrating.
1.) utpreserve ... turned out the be useless, when using vda-confgi -u (it just wipped my data srss data, but did not read the utpreserve file when running vda-config again).
2.) "utuser -o >> file" and "utdesktop -o >> file" nice, did work and saved a lot of my work
3.) vda-backup & vda-restore did work without issues.
4.) did make a text export of all my kiosk "kiosk -e".
4.) lost my utkioskoverride and utresdef settings, so make backups before.
but now I'm back with remote mysql, which seems to be much more stable when running VDI on ESX.
kind regards, thomas

Help on backup & restore

Hi,
We have symmetric configuration with few million mail box per node setup. We are planning to use TSM (Tivoli Storage Management) for backup and restore. The plan is to backup each mount point and restore a specific day or week worth data as needed.
We are not using any message server commands. is it the right approach? Is there any document or white paper other than config guide that explains backup and restore options.
Thanks for your time.
Thanks,
RK 
kotyala wrote:
We have symmetric configuration with few million mail box per node setup. We are planning to use TSM (Tivoli Storage Management) for backup and restore.One issue we hit when migrating to TSM for backups was backing up many small files (which a large messaging server installation has). The number of files resulted in a huge TSM database which caused performance issues for the backup server -- this was 4 years ago so results-may-vary.
The plan is to backup each mount point and restore a specific day or week worth data as needed.Are you planning on backing up at the file-system level and/or block level?
We are not using any message server commands. is it the right approach? Is there any document or white paper other than config guide that explains backup and restore options.The imsbackup utility can be used to create backup files which also contain meta-data about the emails (IMAP flags etc.) Backing up the 'raw' on-disk files means you will lose this information when recovering individual accounts as the data is stored in the messaging server store database.
A common approach for backing up is to use imsbackup to create a large backup file at a temporary location, then use your backup utility (TSM) to backup the imsbackup file. You can use backup groups to perform multiple simultaneous imsbackup's (refer to the manual on how to configure this).
For quick DR recovery you can use block-level backup/restores.
For individual account restores, use the imsbackup image and the imsrestore command.
For both options extensive testing is highly recommended.
There are some backup performance tuning tips here which may also help:
http://msg.wikidoc.info/index.php/Backup_performance_tuning
Regards,
Shane. 
Shane,
Thanks for your quick response...
At this point, our plan is specify required mount names to TSM team and they will start backing up. I am assuming that is file level backup. I am not sure about block level backup. can you shed some light on this please?
Just wondering that are there any benchmark documents from Sun Labs? Please let me know.
-RK 
kotyala wrote:
At this point, our plan is specify required mount names to TSM team and they will start backing up. I am assuming that is file level backup. I am not sure about block level backup. can you shed some light on this please?I suggest you discuss this with your backup group. They will need to explain the concept to you -- this may also not be a facility they offer.
Also you should be considering/testing using the imsbackup utility we provide and getting the backup group to backup the files generated by this command. From the backup-groups perspective, it is better to backup a smaller number of large files (imsbackup output) then a larger number of small files (filesystem level backup).
Just wondering that are there any benchmark documents from Sun Labs? Please let me know.Not that I have seen. We don't offer any-kind of special 'integration' with TSM.
Regards,
Shane.

imsbackup

Hello,
When i use "imsbackup" utility, there is a limit to 2Gb in creating file.
How can i do to make a backup with multiple file or anything else to perform a good backup.
Thank you. 
The workaround is documented. Use stdio:
imsbackup -f -
note the space after the "f" 
If I still want to backup the store to a file, can I use
imsbackup -f - > backupfile? Will there be a file size limitation on this backup file?
Are there any prerequisites to running imsbackup? Should I stop the Messaging server 5.2 first before triggering this command?
Can I run tar instead, while the Messaging server is running?
Is imsbackup -f backupfile the correct command to backup the whole msg-instance/store directory? Or does it only messages in the partition subdirectory? Do I have to put the store instance name in the command?
The posts I posted recently can be only seen when I login. It seems that others can not see my posts. I wonder why?
Thanks. 
If I still want to backup the store to a file, can I
use
imsbackup -f - > backupfile? Will there be a
file size limitation on this backup file?Um, isn't that what I said above? Yes, you can backup the store, no there's no limitation on file size.
>
Are there any prerequisites to running imsbackup?
Should I stop the Messaging server 5.2 first before
triggering this command? imsbackup was designed to be run against a live server. You have no need to stop it.
>
Can I run tar instead, while the Messaging server is
running?Tar on what? the store? Bad Idea. imsbackup was designed to run against a live server. file backup will not get you a "consistant" backup, making restore much more difficult.
>
Is imsbackup -f backupfile the correct command to
backup the whole msg-instance/store directory? Or
does it only messages in the partition subdirectory?
Do I have to put the store instance name in the
command?Please re-read the documentation. it's pretty clear.
If your store is larger than 2 gig, you MUST use the form given above.
imsbackup -f - backupfile /full/path/to/the/store
>
The posts I posted recently can be only seen when
I login. It seems that others can not see my posts. I
wonder why?Not sure what you're talking about. Pretty much nobody besides me will answer, here.
>
Thanks. 
Thanks a lot.
I am an entry lever msg server user. I would like to confirm that the following command ->
imsbackup -f - > backupfile
can backup the full store. There is no need to add "/", or "/mystore" after -f -?
How long would it take to backup a 30GB store?
By the way, how can I check out my store instance name?
Thank you! 
We were also facing same problem. I have one more problem: How to take up PAB's backups using imsbackup. We have found whenever we restore backup taken by imsbackup- user's PAB's are lost.
Any suggestion?
Thanks 
The address book is stored in the DIrectory Server (LDAP). You'll need to backup your directory server. Most people will use the supplied db2ldif and db2bak commands. Search the "Directory Server" forum. This is a common question has been answered in-depth several times.
HTH,
Roger S. 
Answering some questions from precious postings:
"mystore" is shorthand, for the name of your server. You can leave it off..
How do you find out that name? Look at the directory your server is in. msg-<name of your server>. where, "name of your server" is the name that was assigned to the server on install.
And, I agree, PAB (Personal Address Book) is not something that imsbackup, or any method for backing up the mail store will get. It's in the ldap server. 
Thanks. You are very helpful. 
There is still a question here:
our configuration is quite a load balance one: two identical msg servers each on one Sun server use the exact same message store which is attached to the servers as NFS volumes.
I wonder if I run imsbackup on one msg server while the other is still active and uses the same msg store, would this be a problem? Should I stop the other msg server first?
I mean, though, the imsbackup is designed to run on a live server, can it deal with two live servers writing to the same msg-store which are intended to backup?
Thanks. 
There is still a question here:
our configuration is quite a load balance one: two
identical msg servers each on one Sun server use the
exact same message store which is attached to the
servers as NFS volumes. Unless you're using Messaging Server 6.2, you are totally UNSUPPORTED. The use of NFS for mail store was never supported, until 6.2, and is still NOT RECOMMENDED>
Using this kind of setup for load balancing is very un-recommended. Even with 6.2 and NFS, NO TWO SERVERS SHOULD SHARE THE SAME STORE. The system was not designed to do that. Two active servers cannot share the same disk.
If you're talking about HA (High Availability), then that's a different story.
>
I wonder if I run imsbackup on one msg server while
the other is still active and uses the same msg
store, would this be a problem? Should I stop the
other msg server first?Please see above. EAch user's ldap entry includes the name of the mail server. If your servers are the same name, you have an unsupportable configuration, and I highly encourage you to totally re-think your installation, before you have major problems.
>
I mean, though, the imsbackup is designed to run on a
live server, can it deal with two live servers
writing to the same msg-store which are intended to
backup?Nothing about Messaging Server is designed to work in such a situation. Not just imsbackup. NOTHING.
>
Thanks. 
Thanks Jay! It seems I have big trouble here.
There is still a question here:
our configuration is quite a load balance one: two
identical msg servers each on one Sun server usethe
exact same message store which is attached to the
servers as NFS volumes. Unless you're using Messaging Server 6.2, you are
totally UNSUPPORTED. The use of NFS for mail store
was never supported, until 6.2, and is still NOT
RECOMMENDED>Could you please explain more on this? In our case, there are repeating warning messages: NFS write error (10.10.10.10 (the disk array which the NFS volume refers to): Not Owner. We have adopted this configuration for almost one year. Is there particular risks to have NFS volume as our msg store? Again, our msg servers are 5.2.
>
Using this kind of setup for load balancing is very
un-recommended. Even with 6.2 and NFS, NO TWO
SERVERS SHOULD SHARE THE SAME STORE. The system was
not designed to do that. Two active servers cannot
share the same disk.
If you're talking about HA (High Availability), then
that's a different story.Let me clarify our configuration: emails are only directed to mail1 server unless mail1 server is down or users intendly direct their mails to mail2 server. These two msg servers are identical but independent and with different names. Maybe I should call it a HA? So what is the different story?
>>
I wonder if I run imsbackup on one msg serverwhile
the other is still active and uses the same msg
store, would this be a problem? Should I stop the
other msg server first?Please see above. EAch user's ldap entry includes
the name of the mail server. If your servers are the
same name, you have an unsupportable configuration,
and I highly encourage you to totally re-think your
installation, before you have major problems.Our servers have different names. That is why every time before importing mail1's LDAP to mail2, I have to do some editing to make it reflect mail2's name and IP address.
>>
I mean, though, the imsbackup is designed to run ona
live server, can it deal with two live servers
writing to the same msg-store which are intendedto
backup?Nothing about Messaging Server is designed to work in
such a situation. Not just imsbackup. NOTHING.
Thanks.So, at least now, I'd better stop mail2 before do the imsbackup????
Thank you! Your help is deeply appreciated. 
There is more clarification about our configuration:
Only the mailboxes are on the NFS volume and shared by the two mail servers. -
msg-instance/store/mboxlist and msg-instance/store/user are located locally. ONLY msg-instance/store/partition/primary is on the NFS volume and shared between the two mail servers.
So, would this still be a problem? Is this a HA configuration? It seems to me I need to run reconstruct -m on mail2 regularly, under sunch a configuration. right?
Should I stop mail2 before do the imsbackup on mail1?
Thanks. 
YES, it's a problem. You should IMMEDIATELY stop, and re-consider what you're doing.
USING NFS IN ANY SHARED SITUATION IS NOT JUST ASKING FOR A PROBLEM, IT GURANTEES A PROBLEM, SOONER OR LATER. MOST LIKELY SOONER.
You have never stated what version you're using. Unless you're usint 6.2, IN NO WAY WILL NFS WORK.
How can I be more clear?
NFS means, "Not For Store". . . 
Thanks Jay. I am reconsidering my configuration.
I did mention before that the msg server I use is version 5.2.
I have not backuped my msg store for a long time. Before doing anything to the msg server, I want to backup it first. So, what is the best method for me to backup it?
1. Stop mail services on both the meg servers and run tar?
2. Stop mail2, and run imsbackup to backup it?
Thanks.

Backup

I need to take a complete backup of our system to a hard disk which can be used for system recovery. What the safest way to do this. We have OS 9/04, Java Enterprise 2005Q1. This server V250 is our mail server, Directory also on the same server. Its a single server installation.
Thanks in advance 
For one of backups you can run the csbackup command. (see the admin guide for further details.) This will take a current copy of your db.
If you want to take periodic backups then the csstored process can be used to create archival and hot backups. The csstored process is a new process introduced in CS patch 116577-24. The process and it's configuration can be found documented in the JES4 admin guide here:
http://docs.sun.com/app/docs/doc/819-2433/6n4nlfjnq?a=view
If you don't have the csstored process then you should apply a recent CS patch to get it. Here you can configure the csstored process to backup to a different partition/disk from the production db. 
Hi,
There are many approaches to do this - each of which has its own unique problems.
For messaging server and directory server I would suggest you refer to each products respective administration guides which detailed methods of backup.
Each products 'data' can be backed up online in some fashion. imsbackup for backing up emails for mail users; db2ldif to backup the directory server database. This of course doesn't backup application configuration information etc.
To backup a 'complete system' a file-level or block level backup is always an option (makes full restoration in the case of catestrophic system loss a lot easier). Messaging server database doesn't need to be quiesced for backups as in the worst case of a database corruption, it can be deleted and be rebuilt whilst the application is running (although you would want to test/document these steps BEFORE a disaster).
The directory database will of course to attempt to recover from corruption due to not being quiesced, this is where the db2ldif copy of the data can be used to recover lost information. Once again, document/test these steps too.
So a combination of the two approaches would probably provide a safer means of backing up a system.
Hope this information helps.
Cheers,
Shane. 
Thank You Shane for the information.
This will help us to buildt a plan and setup a proper procedure
Regards
Suresh

zpool disaster recovery

Hi,
I have two solaris 10 machines, each one with a zpool DATA.
Now, I should to do a DR of the zpool DATA from the first machine to the
second machine.
If I lose the first zpool, I must have all data saved in the second zpool.
Can I use the snapshot for this purpose? 
The short answer is yes. You can create a snapshot on the source data pool, use "zfs send" to create a stream of the snapshot" and pipe it over using (for example) ssh to the destination machine which can store it using "zfs receive". It is also possible to do incremental transfers. I'd recommend that you check out the documentation (zfs manual page, reference manuals, blogs, etc.) for additional information. I'm not in a position to give you exact references now but perhaps someone else will. 
ok thanks,
but I know that the zfs snapshot is copy on write where you must have the source data for restore the snapshot... and in the second pool I have only the snapshot. Is it wrong?
-------------------------------
of course, I'm writing this thread under 'Solaris Zone' because in DATA there is a zone....
Edited by: 853883 on Sep 19, 2012 12:43 PM 
zpool sharing menas ,ur exporting and importing the pools ?
and where ur taking the snap shot?
while it zfs has snpshot it will not allow you destroy in normal way ,u have to force option to destroy the pool. 
I'm a little confused by your statement. When you create a snapshot, you're effectively saving the state of the filesystem at a particular point in time. When you do a zfs send to send the data to the other machine, it sends the filesystem data as it existed at the time of the snapshot (even if the source filesystem has changed since then). So you do transfer the actual data to the other machine so that the filesystem can be recreated there. 
yes if you use the zfs send command to save the snapshot and moving it to other system and then using zfs receive to restore.
the zfs dataset will be created.
in this case you have to create the snapshots for all the datasets in the respected pool.
then only we can get the complete pool.
Edited by: muvvas on Sep 26, 2012 9:23 PM

Categories

Resources