Jumpstart dependency error: Initialization barrier "mlx4_core" is not provided by any plugin but the following plugins require it: "mlx4_en"

[Impli]

hello,

Installation vsphere 6.0 on a HP DL360 G7 (installation with VM HP-Image)

actual release and HP-driver installed.

because if i take backups of my VM´s with veeam or other Software (always the same effect)  the VM-Host crash down with the following screen.

If i do no backup or snapshot, there is no problem.........     

Anyone an idea ???

[Sharath_BN]

Hi,

Do you use the Mellanox hardware? If yes, what is it configured with. Please also confirm the current version of mlx driver used is as listed on VMware Compatibility Guide - System Search .

The error us caused by the Mellanox driver. Check if it running the latest supported version. If yes, May need to raised with Vendor.

[Impli]

No,

there is no Mellanox hardware! Only Broadcom Network adapter.....

I see the Mellanox driver in the Systemstate/Software components:

I don´t understand this

[unsichtbare]

I just fixed this on our servers. For some reason, the HP customized ESXi ISO was installing the wrong VIB (for us) on BL460 G7 and G9

We could see that the bad VIB (causing jumpstart dependency error) was Mellanox and not VMware by running: esxcli software VIB list

Here's the fix:

esxcli network firewall ruleset set -e true -r httpClient

esxcli software vib remove -n net-mlx4-en

esxcli software vib remove -n net-mlx4-core

esxcli software vib install -n net-mlx4-core -d https://hostupdate.vmware.com/software/VUM/PRODUCTION/main/vmw-depot-index.xml

esxcli software vib install -n net-mlx4-en -d  https://hostupdate.vmware.com/software/VUM/PRODUCTION/main/vmw-depot-index.xml

esxcli network firewall ruleset set -e false -r httpClient

After the fix and a reboot, the Jumpstart dependency error was gone and the installed VIB was by VMware

[tommalt]

Hi,

I'm getting a similar issue with some Mellanox drivers after using the HP ISO to upgrade to ESXi6 U2.

I've tried running the commands to install the VIB from https://hostupdate.vmware.com/software/VUM/PRODUCTION/main/vmw-depot-index.xml‌ but get the following error:

Could not download from depot at https://hostupdate.vmware.com/software/VUM/PRODUCTION/main/vmw-depot-index.xml, skipping (('https://hostupdate.vmware.com/software/VUM/PRODUCTION/main/vmw-depot-index.xml', '', '[Errno 4] IOError: <urlopen error [Errno 97] Address family not supported by protocol>'))

        url = https://hostupdate.vmware.com/software/VUM/PRODUCTION/main/vmw-depot-index.xml

Please refer to the log file for more details.

Would you have any ideas about how to fix this please?

many thanks,

Tom

[mwhities]

The host keeps panicking to this (purple) screen. How were you able to update it otherwise?

[tommalt]

Sorry - in the end we went back to version 5.5 on this host.

[SteveDeWitt]

Any chance you have KB from VMware or HP that says the wrong VIB is being installed? We're having the same problem and this seems to be the fix for it. A reference document would be nice to verify against.

[Donnerwetter]

Hello,

this week we had a simliar failure.

ESX Host with purple screen

main failure:

  JumpStart dependency error: Initialization barrier "mlx4_core" is not provided by any plugin but the followingplugins require it: "mlx4_en"

Server: HP Proliant DL 380 GEN8 ( PSP - Oct 2016)

ESX ISO: HP custom ISO - VMware-ESXi-6.0.0-Update2-3620759-HPE-600.9.6.0.49-Oct2016.iso

last ESX updates and patches from: 26. January 2017

ESX build: 6.0.0 4600944

Opened a Support Ticket and received the following response:

an update for ESX 6.x from January changed the  Interrupt Remapping feature setting  from  default  FALSE  to  TRUE.

This will cause a purple screen with mlx4_core (melanox) driver.

You can check this with the following command:

esxcli system settings kernel list |grep iov

+++++

- Interrupt Remapping has been DISABLED

Option Name      iovDisableIR

Configured Value  TRUE

Runtime Value     TRUE

Default Value       TRUE

+++++

NOTE: The occurance of this failure after appliying the VMware updates could start after some minutes / hours up to , in my case 3 Weeks.

To fix this issue in my case, VMware told me to change the interrupt remapping feature and reboot the ESX host:

esxcli system settings kernel set --setting=iovDisableIR -v FALSE

reboot -d 0

( !! do not use  reboot -f -d xxx  because  -f = force  will not save the changed configuration  !! )

After reboot, you could check the settings again, to be sure everything has been applied successful:

esxcli system settings kernel list |grep iov

+++++

- Interrupt Remapping has been DISABLED

Option Name      iovDisableIR

Configured Value  FALSE

Runtime Value     FALSE

Default Value       TRUE

+++++

Also, we figured out, the melanox drivers were not needed in our case.

So we decided together with VMware-Support, to remove the related melanox drivers.

I did the following:

esxcli software vib remove --vibname=nmlx4-core

esxcli software vib remove --vibname=nmlx4-en

esxcli software vib remove --vibname=net-mlx4-en

esxcli software vib remove --vibname=net-mlx4-core

esxcli software vib remove --vibname=nmlx4-rdma

reboot -d 0

To check if the related drivers were successfully removed, I used the following command:

esxcli software vib list |grep mlx

Best regards and good luck !

Donnerwetter

[d4ngermoose]

This looks like exactly the fix I need but how do we apply this fix if the machine purple screens on boot?

Can't get into SSH session.

[Donnerwetter]

For This situation, you should open a case at VMware Support.

Without direct access to ESXi via SSh in running state, I think, there is no easy way.

Only with booting from other media, then accessing ESXi conf files ..... but that's possibly not supported.

[DeaconZ]

Thank you Donnerwetter! I was having the same issue with some of my HP DL585 Gen7's.

[Moonduza]

I found your response very helpful.

If i don't have mellanox hardware, can i safely uninstall mellanox drivers?