Forum ©XSIBackup: ©VMWare ©ESXi Backup Software
MenuYou are not logged in.
- Topics: Active | Unanswered
#1701 Re: General matters » ERROR (CHKAVRUM) and ERROR (MKROOM01) » 2018-05-27 15:04:03
The [b]df[/b] command is indeed present in ESXi 6.5.0
In fact nothing in your output is telling you about its absence
[root@dusesxisrv02:~] df -h /vmfs/volumes
Error when running esxcli, return status was: 1
Errors:
Cannot open volume:df -h command runs, but it's returning an error. You have some big problem with your datastores setup or your ESXi is plainly broken.
There is other possibility. If you are using one of those evil manufaturer custom distributions, you must know they contain arbitrary modifications, made in sake of the manufacturer commercial interests, and not yours 
, and thus cannot be considered to be ESXi.
If the latter is your case, switch to an official ESXi build ASAP.
#1702 Re: General matters » ERROR (CHKAVRUM) and ERROR (MKROOM01) » 2018-05-26 15:50:15
I guess you mean you are using v. 11.0.0 ("I use the version xsibackup-pro 1.0.0.")
Some user that contacted support lately had a similar issue, it was a hardware problem. That doesn't mean yours is something similar, but most likely in these cases.
Run this command:
df -h /vmfs/volumes/nfs-shares/vmware-vm/xsi-backups/dusesxisrv02And see what is the output.
In your case your path is very deep too: /vmfs/volumes/nfs-shares/vmware-vm/xsi-backups/dusesxisrv02
Try something closer to the NAS root just to make sure it's not the path depth.
NOTE: and do not forget that XSIBACKUP-PRO 11.0.0 RC2 is a [b]Release Candidate[/b] that should not be used in production. We'll release a final version in some days.
#1703 Re: General matters » the VM will not be quiesced, no tools detected » 2018-05-26 07:41:41
Please [b]clicream[/b], read out answers carefully.
Your output clearly states that your VMWare Tools are outdated:
toolsStatus = "toolsOld"#1704 Re: General matters » vmdk not excluded » 2018-05-26 07:39:42
Please [b]clicream[/b], exclude disks as explained in the man page: [url=https://33hops.com/xsibackup-help-man-page.html](c)XSIBackup Classic Man Page[/url]
#1705 Re: General matters » SMTP-TEST » 2018-05-25 14:55:06
Configuring an SMTP server with encryption is not a trivial task, you need to know, not only the basics of the SMTP protocol and its most used extensions, but also OpenSSL, the hierarchy of an SSL certificate and how to configure the SMTP options in regards to checking the integrity of that nested hierarchy. Not to say anti-spam software and setting up filters.
Unless you are going to have some kind of return, setting up and maintaining a full featured SMTP server is not worth the time. Maybe a dummy SMTP with no encryption and a Smart Host relay to be used inside a LAN, like many of our Sysadmin users do, could be worth.
#1706 Re: General matters » vmdk not excluded » 2018-05-25 09:07:36
As per your backup command:
--backup-vms="MTA,MTA2,MAIL1,MAIL2!MAIL1_10.vmdk;MAIL1_11.vmdk;MAIL2_1.vmdk;MAIL2_2.vmdk"You are excluding those disks for the VM [b]MAIL2[/b], not for [b]MAIL1[/b]
You are mixing disks apparently pertaining to MAIL1 and MAIL2 in the exclussion string
#1707 Re: General matters » the VM will not be quiesced, no tools detected » 2018-05-25 08:55:48
This is the command XSIBackup uses to know about the current VMWare Tools status:
vim-cmd vmsvc/get.guest VMId "toolsStatus"Which returns something like this:
(vim.vm.GuestInfo) {
dynamicType = <unset>,
toolsStatus = "toolsNotInstalled",
toolsVersionStatus = "guestToolsNotInstalled",
toolsVersionStatus2 = "guestToolsNotInstalled",
toolsRunningStatus = "guestToolsNotRunning",
toolsVersion = "0",
toolsUpdateStatus = (vim.vm.GuestInfo.ToolsUpdateStatus) null,
guestId = <unset>,
guestFamily = <unset>,
guestFullName = <unset>,
hostName = <unset>,
ipAddress = <unset>,
screen = (vim.vm.GuestInfo.ScreenInfo) {
dynamicType = <unset>,
width = 720,
height = 400,
},
guestState = "notRunning",
powerPolicy = (vim.vm.PowerPolicy) null,
appHeartbeatStatus = "appStatusGray",
appState = "none",
guestOperationsReady = false,
interactiveGuestOperationsReady = false,
}Where "VMId" is the Virtual Machine Id as returned by
vim-cmd vmsvc/getallvmsJust run the above command and check whether the [b]toolsStatus[/b] key shows VMWare Tools are installed and running or not.
#1708 Re: General matters » Remote restore question » 2018-05-24 14:56:58
Read the answer above, we updated information while you were answering.
If the problem persists, just contact support and we'll delve into the problem.
It's nevertheless something about that bug above.
#1709 Re: General matters » Remote restore question » 2018-05-24 14:41:03
Our excuses Jay, you should replace the path "/vmfs/volumes/datastore1", which is just an example, with yours.
I see the problem is you are getting the string config appended to your path.
In the end this is most probably a bug.
Locate this string at aproximately line 253 in the src/pro/restore module (first occurrence of the [b]BTYPE[/b] keyword)
BTYPE=$(eval ssh "$SSHOPTS" -p "$srv2" ${defremusr}@"$srv1" "ls \"${srv3// /\ }\"/config 2>/dev/null \&\& cat ${srv3// /\ }/config 2>/dev/null | grep segments_per_dir | wc -l")Do make sure that you have the string [b]2>/dev/null[/b] right behind
/config, if not add it.
Please note there are two occurences for that [b]/config[/b] string, so make sure the [b]2>/dev/null[/b] code is in each one of those two places.
We fixed this bug some time ago, but you might have an old file.
#1710 Re: General matters » Independent disks cannot be hot backed up » 2018-05-24 14:35:32
Deleting a snapshot does not mean that the data contained in it will be deleted, deleting a snapshot merges data in it into the base disks, but this is common terminology.
So, delete the snapshot, merge the data in the snapshot or reconcile it, and once you have performed that operation, do make sure you can create quiesced snapshots and delete them before even trying to make a backup.
Should your VM become rebelious (that could happen if VMWare Tools cannot handle MariaDB properly) due to heavy load during the snapshot operation, ponder whether performing a non-quiesced snapshot or a warm one. You have many other possibilities before going for non-quiesced or warm in any case. Just delve into how MariaDB handles data and how it interacts with LVM and VMWare Tools during the quiescing process.
#1711 Re: General matters » stuck after sudden storage disconnect » 2018-05-24 09:17:59
If it was already scheduled, you don't need to schedule it again.
Restore the datastore and wait till next cycle, or test manually.
#1712 Re: General matters » Independent disks cannot be hot backed up » 2018-05-24 09:16:14
We don't even know if you are able to take and delete a snapshot so far.
That depends on what kind of backup you want to perform, but as you have doubts about the state of your VM, first thing to do is delete all and make sure you can create and delete snapshots before going any further.
You cannot drive a car if you are not sure whether it has all wheels on or there some missing.
#1713 Re: General matters » Independent disks cannot be hot backed up » 2018-05-23 18:44:55
Then do make sure that you have recomposed your VMX file the way it should be. Being able to run a VM under ESXi does not neccessarily mean that it's correctly configured. You might be running one of the disks on top of a snapshot.
#1714 Re: General matters » Independent disks cannot be hot backed up » 2018-05-23 16:36:57
Please, post the backup command you are using so that we know something more.
Well, as expected its the /bin/vmx process that is holding the file.
The most probable cause is some VM malfunction that 99% of the times is helped by:
1/ Deleting all snapshots and rebooting the VM.
2/ Restarting the host service: # [b]/etc/init.d/hostd restart[/b]
3/ Rebooting the host.
You should take action in the above order until you fix the issue.
If you want to try to dig a bit more into the problem:
1 - Delete all snapshots.
2 - Take a quiesced snapshot (no memory included). If you can't, then there's some issue with the VMWare Tools service, follow advice above.
3 - Inspect the VM configuration while the VM is running and see what are the disk names the VM is running on, they should be:
MAIL2-000001.vmdk
MAIL2_1-000001.vmdk
MAIL2_2-000001.vmdk
MAIL2_3-000001.vmdk
...
We are again guessing, but as the problematic snapshot is named 000002, is it correct to suppouse you are using OneDiff to backup your VM?
If not, you should wonder why you have two snapshots present instead of just the backup snapshot.
#1715 Re: General matters » Independent disks cannot be hot backed up » 2018-05-23 14:51:27
The file cannot be read, that's why you are getting this error.
It cannot be read because some other process is holding it.
The way to find which other process is holding the file is to run this command right before you get that error:
ps -c | grep "$( lsof | grep "MAIL2/MAIL2_3-000002-delta.vmdk" | awk '{print $1}' )"It will most probably be open by the VM process, so make sure VMWare Tools are installed and running and that you meet all the requirements for the system to be quiesced properly. If you are running a Mail server, it will most probably be using a database too. If it's a Windows server, make sure you have all main services configured the right way:
[url]https://33hops.com/troubleshooting-windows-snapshots-in-esxi.html[/url]
The database server might need additional dependent services to be able to quiesce it along with the OS FS.
We have found Windows servers under heavy load (in the middle of a file level backup) to still be problematic to quiesce.
In case you cannot get your OS to be quiesced properly and you can asume some seconds of downtime, you can use [b]warm backup type[/b], which will ensure all files are backed up in a consistent state.
#1716 Re: General matters » SMTP-TEST » 2018-05-23 11:50:15
[url]https://giantdorks.org/alain/fix-for-postfix-untrusted-certificate-tls-error/[/url]
#1717 Re: General matters » SMTP-TEST » 2018-05-23 11:16:11
O.K., that doesn't really matter. Forget the first part of our answer, we will correct it now.
Your problem is strict certificate chain check, which is incompatible with a tipicall ESXi server, especially in the SME world, where very little people will set up a proper certificate structure up to the issuer (Verisign, Thawte, etc...)
So get your mailmaster to remove that check, or swap to a different SMTP server.
#1718 Re: General matters » SMTP-TEST » 2018-05-23 10:20:24
If you are using your own SMTP server, you might need to disable strict full check of certificate chain. XSIBackup's is a self generated key pair. Even if you used the ESXi server's key pair, which is BTW generated on every reboot, it is impossible that the SMTP server can check the full certificate chain up to the issuer's, as there isn't any issuer certificate.
If you are getting this from an ISP, you should call them so that they disable that strict check. Most e-mail users don't have a proper certificate and that check is out of place.
If you are in a corporative environment with strict security measures, you might simply need to just use a different SMTP server. GMail will do it for you.
If you are interested in how certificate chains are built and checked, this is a nice post: [url]https://movingpackets.net/2015/03/16/five-essential-openssl-troubleshooting-commands/[/url]
#1719 Re: General matters » Remote restore question » 2018-05-20 17:29:17
We checked the code and there isn't any bug, we actually remove the last forwardslash in case you add it.
The check that your path is not passing is this:
if [ ! -d /vmfs/volumes/datastore1 ];then echo "Not a directory";else echo "Path is a directory";fiExecute the above piece of code to see what's going on.
There is a remote possibility that the shell expansion is not removing the last forward slash, so remove it first, just in case.
#1720 Re: General matters » seems that the .ERR bug is reappeared » 2018-05-18 14:42:18
We have not designed our software to run multiple instances at the same time for very good reasons.
If you believe you can overcome those reasons, you are free to tweak the code to your needs.
#1721 Re: General matters » possible to backup to 2 places simultaneously or consecutively? » 2018-05-18 11:06:26
Doing it simultaneously would duplicate the used bandwidth and clog on your server. You can do so sequentially, just program two different backup jobs with enough time separation.
#1722 Re: General matters » running incremental every 30 minutes » 2018-05-18 11:00:28
This is something that is at the limits of the "backup" concept and is closer to "distributed FS" or "Real time sync". If on top of that you add an independent disk, please forgive my little sarcasm, but its like presenting yourself to a race with a weight. You want to run fast, but there is something attached to you that is keeping you from reaching your goal.
There's nothing that prevents you from doing it, except the downtimes. If you use a warm backup you will still be unable to backup the independent disk, you need a cold backup for that to be possible. So, I'm afraid that you can't do what you want with your current setup. You have three choices (there may be more):
1 - Change that independent disk to a regular virtual disk, so that you can use OneDiff or some other backup method, like warm backups.
2 - Use some guest level incremental backup to achieve your goal. As said, what you really want is real time FS syncronization, and that falls at the FS level.
3 - Use some distributed FS like [b][url=https://www.gluster.org/]Gluster[/url][/b], [b][url=https://hadoop.apache.org/]Hadoop[/url][/b] or [b][url=https://www.linbit.com/en/products-and-services/drbd/]DRBD[/url][/b] to store your VMs.
#1723 Re: General matters » How to exclude VM's rather then include » 2018-05-17 17:35:51
Thank you for the positive comments and feedback.
We believe the concept can be stretched way further than where we are now. In any case it's becoming harder to find people willing to pay for a "black screen" solution. People in general believe to find more value in a half featured multicolor, semi-transparent, window based application than in more serious software. Virtualization has not only taken over computing science, but people's minds, nobody wants to face the truth in front of an SSH terminal, but just a few brave nerds :-), most prefer the blue pill and click here and there to see if they can get to achieve what they wanted to.
We are nevertheless aware of the fact that we need to make XSIBackup easier to use, which in the end will make it faster and more reliable. Next main release, which we are finishing now, will contain a GUI and will be more integrated with ESXi, will use it's crontab and manage jobs from a file level perspective, allowing to develop new concepts, like distributing full projects in packages and a lot more.
#1724 Re: General matters » How to exclude VM's rather then include » 2018-05-17 16:24:24
[quote=tothand][quote=bpreston]Apologies if you already know this, but if your VM hosts (that you want backed up) are relatively static*, you can send the entire list of VMs (which you [i]do[/i] want backed up) in the commandline (or save it in your VMLIST value and import that).[/quote]
The problem (complication) with this approach is that if you use a shell variable to do this, you can't send this variable to other host with the --host parameter. (Maybe you can, but I haven't figured it out yet)
Probably because the variable name gets to the other host not the content so you would need to maintain a list of VMs to be backed up from all your hosts in one place where the backup cron runs.
Sure, it can be done just as I wrote in my latest post
[quote=tothand]
this complicates things when used with --host because then we need to gather the VM list through SSH from the remote host
[/quote]
The idea to prefix the "important" VMs with some string is nice but if you have a few hosts with a couple of dozen VMs already in place which doesn't follow this naming convention then you have a lot of extra work on your hands to rename them as needed.
(The problem is not the renaming but the whole administrative burden, e.g. the filenames won't match the VM name and if you have some custom scripts using the VM names for something those will have to be modified too.)
Or another approach is to have a static list of the VMs already in place and needed to be backed up AND having a unique prefix for the new VMs needed to be backed up so you end up with a command line like this:
xsibackup ... --backup-vms="vm1, vm2, vm3, REGEX(^prod_)" ...So there are workarounds but none of them offers a clean, easily maintainable, unified solution.
Like you said:
[quote=bpreston]
he "disks to backup" syntax has negative(exclusion) parameters, so it does seem odd that --backup-vms only uses positive(include) parameters.
[/quote]
By the way my problem is not that it can't be done but that no one willing to confirm this from the devteam and say "Sorry, it can't be done right now, maybe we will consider this in a future version" and then I won't have to bang my head against the wall how to do it or why I haven't figured it out yet 
[/quote]
We do like exigent clients and users, but I think you are going faster than we are. XSIBackup is probably the most flexible backup tool in the market, but you are taking the matter to limits that are more in the custom development zone, than in the user program's
#1725 Re: General matters » How to exclude VM's rather then include » 2018-05-17 16:14:55
We are in the middle of big changes, so sorry if we are not offering you the reactivity you expect.
XSIBackup-Pro 11 will change from jobs stored in the xsibackup-cron file to job files stored in the /jobs directory. This will open up new doors to being able to exchange information between hosts in a much more convenient way, and will wipe away all problems related to sending invidual jobs as a string through an SSH tunnel, related to escaping and so on.
On addition XSIBackup-Pro 11 will use the ESXi crontab, so it will be as flexible as a regular Linux crontab is, and much more ;-)
Excluding disks is an excluding feature, while selecting VMs through a REGEXP is an inclusive feature.
In regards to selecting VMs. We are talking about dozens of items, hundreds at most. The simple grep REGEXP functionality should be enough to handle VM selection. This is when having created the VM names in base to an algorithm (something simple, no rocket science) comes in handy. If you have fixed length names with intrinsic meaning, i.e: WINACC001, LINADM002, etc... using a REGEXP is easy. If your naming does not follow a convention, it doesnt matter how powerful your REGEXP tool is, you will be doomed to almost a one by one basis picking.
XSIBackup is flexible enough to offer you many different ways to approach a problem like this.
In any case, please, write to support with your problem's details and we'll give it a thought and publish it here for others to take advantage too.