freeze on boot

[Rudym12]

Good day all,

Having an issue running Zabbix on Hyper-V, keeps freezing on "ransom" 7 urandom warning (s) missed due to retelimiting"
Have tried running on 2008, 2012 server, and windows 10 Pro all with the same result.

BTW, i tried in two different network locations.

I have tried looking for a solution online but can't find any info. any help is greatly appreciated.

Thanks
RudyM

[provels]

I had a similar problem on 2012R2. I ended up using the 4.4.6 VHD image available at https://sourceforge.net/projects/zabbix/files/ZABBIX%20Latest%20Stable/.
Used Gen 1, 2GB RAM. Hope this helps.

[tim.mooney]

If you search online for the exact error message (what you provided has a couple of typos in it), you'll see many, many people reporting the same issue during boot, for different Linux distros. This isn't a problem that's specific to Zabbix, but unfortunately it appears to be specific to the kernel used by the Linux distro in the appliance.

Based on reading a few of the posts from people that have seen this in other situations, it seems to be that the entropy pool is not being filled fast enough at boot time, and parts of the boot process that require random numbers are exhausting the pool. The Linux kernel uses multiple sources of unpredictable events as input to the entropy pool, and that in turn helps the kernel's pseudo random number generator (PRNG) provide random numbers to cryptographic software that needs it. When the pool hits empty, anything that uses /dev/random is going to block (stop) until there's enough entropy in the pool to fulfill the request.

There are many things that the kernel uses to fill the entropy pool, but timings for key presses and mouse movements are two things that are easiest for a user to provide. It's not clear to me from the posts I looked at whether the pool being exhausted is because of a kernel bug or some other issue, but I saw several posts that indicated that doing things to provide more entropy helped avoid the problem.

To debug whether you're having the exact same issue (I suspect you are), try this:

1. reboot the Zabbix appliance
2. as soon as the kernel boot starts (after the initial timeout or welcome screen), start pressing keys on the keyboard quickly. I know that sounds bizarre, but I'm not joking: the timing between keypresses can help with entropy gathering, and the more key presses the better. To avoid pressing something that's going to be interpreted by the eventual login prompt or GUI dialog, you can just repeatedly press the Shift key(s) on your keyboard. The more keypresses, the better.
3. If you have a mouse attached you can move that back and forth quickly too. Mouse movements contribute to entropy gathering (even before the GUI starts up at boot time, the device input is being registered) as long as they're in the window for the VM. Mouse movements outside that window probably won't be passed through to the appliance.
4. If you do this for long enough, the entropy pool will accumulate enough entropy that the random # generator should be able to provide the randomness that's needed for cryptographic calls used during the boot process.

If this gets you to the point where the appliance boots and everything starts up, report back.

Once the appliance is up, it may be possible to do something that will avoid the block waiting for entropy, so you don't have to pound on the keyboard like a maniac during boot. If I were in your situation, I would first see if there's a kernel update available for the kernel in the appliance (that's pretty likely where the issue is). Barring a kernel update that fixes the issue, the next thing I would look into is whether there isa software package like "egd/haveged" that can be used to supplement the kernel's entropy gathering routines.

[provels]

I'm sure they tested these images before they uploaded them to Sourceforge, no?

[pzabortsev]

Zabbix appliance 5.0.0 for Hyper-V have the same issue
It is very strange. Is it really nobody of Zabbix team read this messages?

[Old Pete]

Hi guys, I have exactly the same problem using Hyper-V and Zabbix appliance 5. Somebody went to a great deal of trouble to make this appliance, but I can't use it because of this problem. Tried on our Production Cluster and also on Hyper-V on windows 10 with same issue.