3

On two Proxmox KVM VMs, I get the error hcp: ERROR: FALLOCATE FAILED! in the console. I get this error on two separate Ubuntu 16.04 VMs.

I am not sure what the error means, or where it comes from. I should note that I saw the error in both consoles after ejecting the installation CD in Proxmox. I re-attached the CD to both VMs just to be sure, but saw the error pop up in the console of one of the VMs again just now.

The VM itself seems to function. I found a thread on the Proxmox forums: Corrupt Filesystem after snapshot, but I do not have any snapshots created for either VMs.

I am confused because I have no idea where the error comes from. Unfortunately, I'm not sure if it appeared in the console before I ejected the installation CDs as well.

I am using R1Soft which loads an hcpdriver. Perhaps that's related (though I wouldn't have any idea why it would show these errors).

df -h output on both servers as requested:

Screenshot 1 Screenshot 2

William
  • 163

1 Answers1

3

What causes this error is pretty straightforward. The error is caused by the fallocate command failing and returning with an exit code of 1 (fail) as opposed to succeeding and returning an exit code of 0.

The error in and of itself does not effect the system, it is actually a symptom of an underlying problem.

The root cause could be an attempt to allocate more space for a file than the system has available or perhaps a corrupted filesystem (whether actual or virtual disk). I would expect the full error in the first case to be fallocate failed: No space left on device

A failure to deallocate space could also throw an error. This IMHO would point toward the aforementioned corrupt filesystem.

If you do not have a current backup/image/clone of the filesystem in question it would be prudent to make one. Doing so would ensure that you could return to where you are now without having to be concerned about the unintended consequences of rebooting the VM. If the underlying disk hosting the filesystem is damaged you may wish to use something like ddrescue in an attempt to preserve as much as possible from the physical disk. This message will not cause a failure to reboot the VM however do not take that statement as a guarantee that your VM will reboot successfully. Anything can go terribly wrong at any time and a detailed view of the future is not something I (or anyone) can provide for you.

For more detail on the fallocate command see man fallocate

EDIT: It's entirely likely that your R1Soft backups are using snapshots without your knowledge. See this which seems to increase the relevancy of the link you posted in your question. It would seem to be worthwhile to check the version of the driver/agent installed on the system to ascertain if you have the latest version and whether any bugs have been squashed that would indicate whether you should update it or report a bug.You might consider exploring alternative backup methods in the interim for safety if at all possible.

Elder Geek
  • 36,023
  • 25
  • 98
  • 183
  • I am certain that the physical disk is fine. I'm wondering if the system would survive a reboot. – William Jan 25 '18 at 17:16
  • 1
    create a current backup and you won't have to wonder. – Elder Geek Jan 25 '18 at 17:21
  • You could also clone the VM – Elder Geek Jan 25 '18 at 17:25
  • I have a backup. Just wondering if the current VM in its current state would survive a reboot. – William Jan 25 '18 at 19:44
  • 2
    Even after 30 + years of experience I am unable to tell the future. Anything can fail for any reason at any time. This is why we backup. Probably is the best I can give you. – Elder Geek Jan 25 '18 at 23:17
  • I added screenshots of df -h output as requested. – William Feb 01 '18 at 18:05
  • I see nothing glaring in the snapshots that indicate anything in particular. It doesn't appear that fallocate is failing due to a current lack of storage space. I suppose that depending on your use case you could be running out of inodes. df -i will tell you that. You say you have a backup and no snapshots. How and what did you backup? – Elder Geek Feb 01 '18 at 18:50
  • We checked. Too few inodes doesn't seem to be the case. – William Feb 02 '18 at 17:00
  • We're doing backups with R1Soft and I'm fairly sure that's why those messages are getting dumped. R1Soft loads a hcpdriver and other Ubuntu VMs without that driver / R1Soft backups don't get the notice. – William Feb 02 '18 at 17:01
  • @WilliamD.Edwards updated answer – Elder Geek Feb 02 '18 at 21:02
  • 1
    "It's entirely likely that your R1Soft backups are using snapshots without your knowledge." R1Soft uses snapshots and it is to my knowledge. The CDP agent driver seems to be incompatible with the kernel version, due to an error in R1Soft's documentation. I rebooted both VMs and they seem to have survived the reboot. So, R1Soft is very likely to blame here. – William Feb 05 '18 at 20:47
  • @WilliamEdwards, you should write an answer tou your own question with that information, for other people to find, or at least add an update note to your own answer. – MV. Jun 15 '18 at 20:29
  • @MV. I don't have an answer. – William Jun 16 '18 at 10:59
  • @MV William is quoting my answer last edited 2/2/18 – Elder Geek Jun 17 '18 at 14:43