ghettoVCB.sh - Free alternative for backing up VM's for ESX(i) 3.5, 4.x, 5.x, 6.x & 7.x

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

lamw · ‎03-30-2010

What I would like to see right now from the community are ideas/features that you would like to have, based on the feedback and the amount that I receive, we can go about figuring out whether they would be feasible.

If certain feature requests have already been created/hacked together, more than welcome to take a look but I would like that to be kept on a separate thread so that when troubleshooting issues, users are using the current version of the script and not other various/modified versions.

As of now I've seen the following:

Ability to power off VM, offline snapshot and power the VM back on and proceed with the backup.
Global ghettoVCB configuration file
Implementation of some locking mechanism to ensure only 1 ghettoVCB instance can run at any given moment
Support VMs with existing snapshots
Ability to control whether VM backup directories are rotated in general or kept in-definite and leave it to end users to rotate out

I'll continue to update this list as I receive more feedback.

I most likely will be conducting a private BETA for a select few as I've done in the past for previous releases, for those that are interested and has a significantly sized environment to test various functionality/features, shoot me an PM with your Name, Email and Quick summary of the testing environment (ESX(i) version/build, # of VMs, size of VMs, etc)

Thanks again for all the great feedback

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

scowse · ‎03-30-2010

I can confirm that there is a limit on crond which results in evaporation of settings in esxi.

This is why I now use plink from windows

P.S. To all you people out there who, like myself, have been going through the veeam, nfs, scp, woefully slow transfer speeds nightmare - scrap it all!

Centos 5.3 x64 iscsi datastore with 1Gbit nic and ghetto = 100 MB/sec (yes - that is 100 megabytes per second:)

Six commands to configure the target and six clicks to configure the datastore.

If everyone did this the current page would be half its length ;:

gautelund · ‎03-31-2010

Interesting. I could reproduce this once, but after that I tried to add data beyond 512 bytes to the crontab several times, using vi or redirecting, with no problems.

Well, what the hell - 512 bytes ought to be enough for anybody.

imorgan · ‎03-31-2010

You say " one way is just to fully disable this OR by setting a very high count if you don't want any deletion."

Well, I do not want the script to ever rename a past backup directory, which suggests that my only option is to "fully disable" rotation. But how do I do that? I have read the whole script, and can find no means to disable rotation entirely, and clearly setting VM_BACKUP_ROTATION_COUNT=0 would be bad indeed.

Here's the deal: I need to rsync the backups off-site over a slow link. Since ghetoVCB renames the directories when rotating, ALL the backups have to be re-sent, not just the most recent. Can't the script just keep the most recent N backups directories? Why does it have to maintain a rotation count in the directory name at all?

If I could just get ghetoVCB to stop adding a rotation count to the directory, or disable rotation entirely, then I can apply my own rotation logic outside of ghetoVCB, and then my rsync's will work nicely.

Thanks for ghetoVCB. Until I started moving backups off-site, it worked like a charm! Any insight you can provide to my problem would be greatly appreciated.

lamw · ‎03-31-2010

When I say fully disable, you just need to comment out the function call in which the rotation is checked. If you do 1 backup per day, this should be fine. Though if you do multiple backups per day, the default VM backup directory is not unique enough as it only holds "YYYY-MM-DD" and you will most definitely override any earlier backups.

The whole point of the rename is similar to how system logs are rotated, you want to be able to keep track if you have multiple backups per day to see which is the oldest so you can purge after certain count, hence the additional number is appended in the form of "--X". If you want to suggest an alternative way for this, by all means I'm open to suggestions. This has never been a concern or problem for others, but if it is, I'm always up for improvements. I just need to understand the use case and how a change may affect others/etc.

I can look into setting a configuration param that allows the user to decide whether rotation will be done by the script, but again, there needs to be some uniqueness being created on a per backup basis, else you'll override the previous backup. Perhaps include additional timestamp information (ss:ms) which would help and then maybe purge after n-copies?

Thanks

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

goppi · ‎03-31-2010

@imorgan + lamw:

We modified the script for exactly the same reason.

As we rsync the ghettoVCB backup folders to external storage, the folder names of the backups must not change, otherwise the same backup folder would be rsynced several times after a rotation took place.

We changed the skrpit in the way that a a backup folder only carries the current date and would not change for rotation. IMHO there is no reason for changing names on rotation. The folder name itslef (the date) is enough to find out which folders are the oldest ones for beeing deleted when rotation takes place.

We do not have the case of multiple backups on the same day which is not covered by this naming scheme, but not a problem anyway as you could simply add a continous number if a folder with the current date already exisits when having multiple backups on the same day.

Regards,

grubi.

lamw · ‎03-31-2010

So from what I'm hearing, it sounds like for rotation:

1) Flag to disable

2) Flag to enable

In both cases, have some logic to figure out what the next number would be in case user performs multiple backups per day? Or should we screw the numbering and just add on "ss:ms" to ensure uniqueness and then just purge after N where N = max to keep? if we're in scenario #2?

Is that what I'm hearing, that way it won't break things like rsync and probably less work for the rotation function since you don't need to rename directories. Originally I was using that to keep track of the older copies and a way to extract the current count.

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

deniseche · ‎04-01-2010

Hello,

Thanks for this work ! It almost work for me but i have a strange issue.

Here is my environement :

Host server : Esxi 4.0 Update 1 (last version)

One virtual Machine, Size : 140Gb

Backup datastore type : NFS. Hosted by a nas appliance (iomega).

Here is the issue. The first backup is taking place, Vm is backuped up on the nas. But when i have a look on the log the last availbale line is --- 2010-03-27 10:15:46 -- info: Removing snapshot from server.mydomain.com ...

Next backup I have this error :

2010-03-29 00:19:02 -- info: Snapshot found for server.mydomain.com, backup will not take place

The intersting thing is no snapshot on the vm when i use vpshere client. If i take a snapshot on the virtual machine a new snapshot appear called : Consolidate Helper - 0

If i remove all snapshot nextime, the backup work but i have the same issue.

Here is a link to my logfile http://pastebin.com/nq9NPNU4

The problem seems to be in snapshot removal, is there something to do ?

Thanks for your help.

Denis.

lamw · ‎04-01-2010

You're affected by the "NFS" problem, take a look at FAQ #18

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

romgo75 · ‎04-01-2010

Hi,

I'm running with lot of troubles ghettoVCB.

I started from scratch, since I had trouble before. I want to backup 3 VM. I'm using a backup config file for one of my VM.

0 23 * * * /vmfs/volumes/datastore1/vmBackup.sh -d debug -f /vmfs/volumes/datastore1/vm_backup_list > /vmfs/volumes/datastore1/ghettoVCB-backup-$(date +\%F).log -c /vmfs/volumes/datastore1/backup_config/

But the first run of this backup only produce a backup for one VM and it didn't compress it.

any idea ?

lamw · ‎04-01-2010

Hello,

You apparently did not go through the documentation and read the FAQ's, specifically #0 and #1 and since you're using compression, I would also recommend taking a good luck at the compression FAQ's.

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

rdhoore108 · ‎04-01-2010

Hi William,

I have a humble suggestion: to add a command line parameter where we can tell the script NOT to remove previous backups of a VM if there was any error condition during the current backup of this VM, but to remove the current failed one instead.

Imagine I have just enough disk space for one complete set of all my VM's, and for making a new set (before the previous is deleted). So my VM_BACKUP_ROTATION_COUNT = 1. Now, if "clone disk" fails for a VM for any reason, I end up with an incomplete new backup, plus my previous good backup is gone (deleted by the script).

As it is now, the script joyfully declares the backup successful, and removes my previous (truly successful) backup. For example:

(...)

Cloning disk '/vmfs/volumes/Raid5/SQLSRV1/SQLSRV1_1.vmdk'...

Clone: 49% done.Failed to clone disk : No space left on device (1835017).

2010-04-01 17:22:00 -- info: Removing snapshot from SQLSRV1 ...

2010-04-01 17:22:05 -- info: Backup Duration: 18.18 Minutes

2010-04-01 17:22:05 -- info: Successfully completed backup for SQLSRV1!

Maybe the script could look for an error condition, and consider the operation failed instead of successful, retaining the previous backups and removing the last instead (when a parameter of your choice is passed, for backward compatibility).

Thank you so much for considering this. My sincere apologies if this has been suggested before, I've tried reading through all comments, but gave up halfway, there are just too many.

Regards,

Ronny

lamw · ‎04-01-2010

Yes this has been suggested already, take a look at my comment on Mar 30, 2010 1:25 PM

I'll still need to figure out all the logic and the various use cases, but yes, this is something I'll be considering.

Thanks

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

ianwitherow · ‎04-02-2010

Hey,

I'm getting an error no matter what VM I try to back up. I'm running the command:

./ghettoVCB.sh -f

(file name has one line, the name of the VM). No matter which I try, I keep getting:

"2010-04-02 21:48:18 -- info: Error: failed to locate and extract VM_ID for FTPServer!"

(FTPServer was the example)

Is there a line I should be changing to tell it where the VMs actually are on the SAN array?

lamw · ‎04-02-2010

So if you're positive that the VM's displayName is in fact "FTPServer", what you can do to troubleshoot is the following command:

ESX:

vmware-vim-cmd vmsvc/getallvms | grep -i "FTPServer"

ESXi

vim-cmd vmsvc/getallvms | grep -i "FTPServer"

Something should return back with a line that specifies the VM's vmid along with other attributes if in fact this VM exists.

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

ianwitherow · ‎04-02-2010

Thanks for that reply, it was very helpful.

It's showing me the VMs from one of the folders we have set up. Is there a fairly simple way to increase the scope of that command to include other folders where we have VMs?

eg: "/vmfs/volumes/ESXi-VMs" as well as "/vmfs/volumes/A2VD1_VMs"

Thanks

lamw · ‎04-02-2010

So the command that I'm showing you from above, lists only REGISTERED VMs on the current ESX(i) host. If it's not being displayed, then it means either A) VM is not registered B) it's running on another host.

I'm basically providing you the method in the which the script figures out whether or not the VM you're looking to backup exists to pull out it's VMID.

In your case, it sounds like it's not running on that host or you're getting the displayName wrong.

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

goppi · ‎04-03-2010

@ianwitherow :

You also will encounter this problem if the file containing the machines to backup uses the wrong EOL characters.

This is somewhat difficult to diagnose as the name of the VM displayed looks corret.

Please ensure that the mentioned file is created under Linux and not under Windows or check with a hex editor that you don't see a CR+LF at the end of each line.

HTH

goppi.

deniseche · ‎04-05-2010

I had a look at FAQ #18, but i don't have Input/Output error in my log...

By the way i tried to change from NFS protocol to Iscsi and i have the same issue.

Any idea for a solution ?

Regards.

Denis.

lamw · ‎04-06-2010

Well the input/output is one of many errors you may or may not see, basically from what you're describing, you're seeing similar problems when trying to delete larger files which also has a side affect of snapshots potentially being left behind. FAQ #29 also is related to this problem and it's not the NFS protocol specifically but it's still something we're not 100% sure. Though if you're switching between NFS and iSCSI and you're still seeing the error, then it looks like FAQ #29 may apply to you. Try switching the underlying filesystem type which I'm assuming you're using ext3 to xfs and try it again.

You'll also notice you won't run into this problem if you backup to a local VMFS volume on the host, so this specific to the datastore and it's configurations.

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

Miriano · ‎04-08-2010

Hello William

I thank you for the excellent script that makes life easier for us system administrators like me!

if the information can be useful to other friends, I have scheduled the script from a Windows machine, and in combination with WinSCP and Plink BLAT I got a solution that makes me sleep peacefully!

the real limit is that sometimes I find infrastructure machines with old snapshots.

you think you can, provide these features in the script:

- Shut down/Suspend(best) only one specific VM has a snapshot without make a new snapshot..

- Perform a complete copy of the total of all files related to the VM in the path of the backup

- Restart the same VM as if nothing had happened!

sorry if any argument was already addressed, but I tried to read the whole discussion and I found nothing.

thanks again for your work!

hello

Miriano.

ScottChapman · ‎04-15-2010

It would be nice to see a mechanism to run a command on the guest before and after the snapshot was taken.

My VM happens to be an email server, and while I probably can just have the whole thing shutdown, it might be "cleaner" just to have the guest prepare for the snapshot by shutting down any running services.

Could that be easily added?

lamw · ‎04-15-2010

No, this is not something that can be added. The ability to perform operations would differ from OS to OS and there's nothing you can leverage through a "free" solution to perform any type of operations within the guestOS. If you have VMware Tools installed, and you enable quiesce of the filesystem, it'll flush the IO to disk before the snapshot is taken, certain OSes such as Windows may have some built in frameworks as VSS to make the application aware that it's about to backed up. In either case, the snapshot should not affect the application running and it'll continue to perform r/w operations but all new writes will just go to a new delta disk.

You of course could add your own "scripts" to say shutdown a service/etc. for instance SSH into a UNIX/Linux system but again this will be environment specific and is outside of the scope of the script.

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

ScottChapman · ‎04-15-2010

I am not so concerned about eh VM continuing to run, but rather the usefulness of the snapshot that is taken.

I think having a CMD that is run locally on the host before and after the snapshot would give me the ability to write a guest specific script that could essentially perform an SHH to the guest to run a script there that would do my shutdown and restart.

I can add something to the existing script, but it seems like it might be a generally useful hook to add for other reasons...

N'est ce pas?

lamw · ‎04-15-2010

I am not so concerned about eh VM continuing to run, but rather the usefulness of the snapshot that is taken.

Not sure I follow your comment, the "usefulness" of the snapshot is that it's required for powered on VMs to unlock the VMDK for read access, else it'll be locked. So that's a requirement, even for other backup solutions such as VMware vDR and Veeam Backup & Replication.

If you have a need to run guest specific commands/scripts before and after the backup, this is something that you'll need to create for your environment. Depending on restrictions/requirements of a given environment, the implementation of this can be done in various ways (e.g. using SSH keys,etc).

Again, this is outside of the scope of this backup script, the point of the script is to provide a general backup script that can be implemented in any environment. If you need more specific operations that applies to your operating environment, then you'll need to implement those features yourself.

Thanks

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

ScottChapman · ‎04-15-2010

Like I said, my machine happens to be an email server. So, just grabbing a snapshot while it is processing email might not be great. For example (purely hypothetical), imagine it is in the middle of handling and incoming email so, pipes are open reading streams, and the DB has a transaction open, etc... Restoring to that exact point might not be useful since the incoming port will no longer have the data from the email server that sent the email in the first place (it would have been consumed by the continuation of the running VM).

So, ideally I would shut down the email services gracefully prior to performing the backup to make sure the machine is nice and "quiet" for the snapshot to more likely be useful.

At any rate, I am not looking for the scripts to actually do any of this, I can write that. I am just suggesting that having a hook in your script that allowed someone like me to specify in the VM's backup config file, a script that should be run prior to taking the snap shot, and another one after the snapshot was taken would be ideal. Something like:

PRE_SNAPSHOT_CMD='shutdown.sh'

POST_SNAPSHOT_CMD='restart.sh'

And ghetto would simple (if the strings were not blank) execute those scripts on the host (checking the return values of course).

Then I could write the shutdown.sh and restart.sh script to do what I need for that host running against my specific guest OS.

See what I am getting at?

lamw · ‎04-15-2010

I see where you're coming from but again I'm going to re-iterate my points. This is "specific" to YOUR environment, not everyone. I'm not going to make general assumption on how the script should be executed. What if I have some users that can utilize something like SSH keys to do the SSH to the host to execute the script and others that can't? What then? What if I have some users that are running Window OSes and not Linux OSes? You won't be executing the script in the same manner using a shell script/etc. What about isolated network management and VM networks? What if a user wants to run script A,B,C on VM1,3,4 but not 2 and 5? The logic to implement something like this definitely not trivial and can be very complicate or very simple depending on the environment and the use cases.

Regarding feature requests, I would like to point you to FAQ #15

The best advice/suggestion that I can offer is you implement something that fits your needs, I will not be creating any type of generic solution for this as there are too many variables / assumptions to consider.

Thanks for the feedback.

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

ScottChapman · ‎04-15-2010

That's the whole point of just providing the hook; it is up to the user to write the actual script that is appropriate for their environment. The only assumption that ghetto would make is that if it is set, and points to a valid file, it would perform a system call to it and get the return value; erroring out on a non-zero return value. And that would be the extent of the contract ghetto would have; everything else would be up to the user.

Also, because it would be in the VM config file, they could specify different scripts for different VMs if they wanted to.

If the user needed to run script A,B,C for VM1 they would write a script that ran A,B,C and specified that script in the parameter in the config file for VM1.

I understand your skepticism that this would indeed be useful to others, and thus not really interested in implementing it. I would be more than happy to implement it and give it back to your for your consideration.

lamw · ‎04-15-2010

Like I said, there's definitely a way to implement something like this but it will not be a trivial task. This would potentially add an additional level of complexity on top of the script and additional testing. My main concern is what if these "user" scripts never return? Debugging whether it's the script or user scripts? Creating a simple Linux scriptp and using something like SSH keys wouldn't be too bad, but what about the use case for Window base systems? Creating the "user" script will differ between ESX and ESXi, since Busybox console will be much more limited. Simple SCP would work for UNIX/Linux systems but to get a script onto Windows box may not be as trivial and may require other binaries/components.

If you come up with a method by all means, though this is something I'm not going to be adding into the script nor supporting.

Good Luck

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

Vishy1 · ‎04-18-2010

Hi ALL,

I have a small issue where I have in the global file VM name “WinXP Admin” with a space in between doesn’t get reflected to /tmp/list file during comparison. However in the /tmp/currentvms file, it does pickup that VM name. Can someone see what I am doing wrong here.

start

rm -rf /tmp/list

vmware-vim-cmd vmsvc/getallvms | sed 's/[:blank:]\{3,\}/ /g' | awk -F' ' '{print $2}' | sed 's/] /]\";\"/g' | sed '1,1d' > /tmp/currentvms

for i in `cat /tmp/currentvms` ;do

awk -v var=$i -F, '$1==var' /vmfs/volumes/BACKUP/scripts/globaltobecloned >> /tmp/list

done

/vmfs/volumes/BACKUP/scripts/ghettoVCB.sh -f /tmp/list

Thanks in advance.

Vishal

If you found this information useful, please consider awarding points for Correct or Helpful.

webifi · ‎04-18-2010

I see as a listed feature: "Ability to include/exclude specific VMDK(s) per VM".

However, after looking through the documentation, I see no specific language on how to "exclude" a VMDK, only how to specifically include a list of VMDKs, or "all". This implies that the method for exclusion is to simply exclude a VMDK from the include list (include all but the one you don't want.)

I would like to completely exclude a specific VM, and all of its VMDKs from being backed up. How do i accomplish this? Do I simply set VMDK_FILES_TO_BACKUP="" for that VM? Is there another way to tell ghettoVCB to skip the VM in its entirety?

Thanks,

--John

lamw · ‎04-18-2010

This is an easy one

By default, not all VMs get backed up. As you can see from the documentation, you need to specify a vmlist which tells the script which VMs to back up, so if you want to skip a specific VM all together, just don't include it. If you need a more granular backup config such as specifying specific VMDK, then you need to create a backup configuration file for each VM in which you can specify exactly which VMDK(s) you want backed up. Again, all this relies on the vmlist and the VMs within the file. Hopefully this makes sense, if not, I would recommend you go through the documentation again.

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

lamw · ‎04-18-2010

Hello everyone,

I've been spending some time looking at the suggestions/features provided by the community and have the following listed:

1) Ability to backup all VMs w/o having to specify a list
2) Ability to power off VM, offline snapshot and power the VM back on and proceed with the backup.
3) Global ghettoVCB configuration file
4) Implementation of some locking mechanism to ensure only 1 ghettoVCB instance can run at any given moment
5) Support VMs with existing snapshots
6) Ability to control whether VM backup directories are rotated in general or kept in-definite and leave it to end users to rotate out
7) Additional logging + overall final status of backup job

As of writing this, I've completed the implementation of #1, #3, #4, #6 and #7. I've still got a bit more of testing to do but will be releasing an early preview of the script for the general public.

I will not be implementing #5 as snapshots should not be used for long durations but also for technical reasons as the implementation of removing snapshots by vimsh is not the most robust and could lead into issues especially with complicated snapshot trees. With regards to feature suggestion #2, I still need to explore what changes that may mean for the script as it may not be a straight forward change.

Again, thanks to the community for all the great feedback and to the users who have written some enhancement to the scripts, I've taken some of features and integrated into the new script.

Keep an eye out for the new version of ghettoVCB very soon!

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

romgo75 · ‎04-19-2010

Hello,

I have got some issue with the script.

here is my log file : http://pastebin.com/NN5BasUw

ESXI 4.0

SMP Release build-219382

vm1 : vmdk 20 Go

vm2 : vmdk 20 Go

Backup datastore type : local

In my backup folder I have got the following folders :

vm1

vm2

Inside the folder vm2 :

vm2-2010-04-19 vm2-2010-04-19~~2 vm2-2010-04-19~~3

Inside the folder vm1 :

vm1-2010-04-19--3

I try this script on two different server (running the same esxi version). I was running the backup with the crontab.

When I'm running the backup manually, everything seems allright.

Thanks in advance for any help.

In my vm2 folder I still have the snapshot (delta files are still there).

Any idea ?

romgo75 · ‎04-19-2010

Sorry I forgot to give the capacity of the datastore 2 :

1.8T 642.6G 1.2T 35%

lamw · ‎04-19-2010

Quickly looking at your logs, it seems that you have multiple backups running at the same time, especially vm1. I would double check your crontab entry to ensure you don't have multiple ghettoVCB instances running which may be over-stepping each other and that maybe why you're still seeing delta files.

If you can manually run it without any problems, then it should be the same if you have it scheduled via cron. In the next release, a check has been implemented to ensure that only 1 instance of ghettoVCB is running at anytime on a given host.

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators

If you find this information useful, please award points for "correct" or "helpful".

romgo75 · ‎04-20-2010

Hello,

thank you for you answer. I told you a bit quickly I didn't had any trouble by manual execution, but I finished my test and I have exactly the same issue.

I try to run the script on my other ESXI server which is exactly the same configuration and I have trouble too.

I used the following crontab :

01 * * * * /vmfs/volumes/datastore2/ghettoVCB/ghettoVCB.sh -d debug -f /vmfs/volumes/datastore2/ghettoVCB/vm_list -l /vmfs/volumes/datastore2/ghettoVCB/backup.log

Regards

lamw · ‎04-20-2010

Just released the latest version with the script with the changes I had mentioned earlier, please give the script a try and let me know what you think or if you find any issues/bugs.

Thanks and enjoy!

=========================================================================

William Lam

VMware vExpert 2009

VMware ESX/ESXi scripts and resources at: http://engineering.ucsb.edu/~duonglt/vmware/

VMware Code Central - Scripts/Sample code for Developers and Administrators