VMware Cloud Community
MihirP
Enthusiast
Enthusiast
Jump to solution

Host Disconnection

Hello,

Couple of our ESXi 5.0 hosts had got disconnected and withing few seconds showed connected in vSphere client.

During this few seconds, VMs on it also alerted disconnection.

I could not check whether host could be connected directly in this few seconds because I was not in shift.

I found one article VMware KB: VMware ESXi 5.0/5.1 host fails to reconnect to VMware vCenter Server due to vpxa coredump, which states to change the "the thread stack size".

--> I want to know what is this "Thread Stack Size" ?

--> What is the cause of below showing in vmkernal.log;

2015-06-26T10:28:19.344Z cpu2:15682393)WARNING: UserObj: 675: Failed to crossdup fd 8, /vmfs/devices/char/vmkdriver/hp type CHAR: Opaque service console status
2015-06-26T10:32:28.378Z cpu2:15682393)WARNING: UserObj: 675: Failed to crossdup fd 3, /vmfs/devices/char/vmkdriver/hp type CHAR: Opaque service console status
2015-06-26T10:32:28.378Z cpu2:15682393)WARNING: UserObj: 675: Failed to crossdup fd 8, /vmfs/devices/char/vmkdriver/hp type CHAR: Opaque service console status
2015-06-26T10:36:37.389Z cpu2:15682393)WARNING: UserObj: 675: Failed to crossdup fd 3, /vmfs/devices/char/vmkdriver/hp type CHAR: Opaque service console status

Thanks.

0 Kudos
1 Solution

Accepted Solutions
gallycool
Enthusiast
Enthusiast
Jump to solution

Hello Mihir,

If the host is having more number of snapshots for the virtual machines you will be running in to such problems.

Please heck on the virtual machine for large number of snapshots.

The default value of ThreadstackSizekb is 128 and can handle only limited number of snapshots.

If you exceed that number then the host will e disconnected.

Instead of increasing the threadstack value we can delete or consolidate snapshots for virtual machines.

There is a line in the vpxa.cfg file on the host that has an entry <ThreadStackSizeKb> and the value is set to 128.  When vpxa service attempts to connect to vCenter this Thread Stack is limited to not allow VMs that are too many snapshots.


If you don't want the snapshots to be deleted we will increase the threadstacksize so the host can handle more number of snapshtos.


Please let me know if you still have any further queries.


Thanks

Sam





View solution in original post

0 Kudos
3 Replies
MihirP
Enthusiast
Enthusiast
Jump to solution

Awaiting reply from someone please.

Thanks.

0 Kudos
gallycool
Enthusiast
Enthusiast
Jump to solution

Hello Mihir,

If the host is having more number of snapshots for the virtual machines you will be running in to such problems.

Please heck on the virtual machine for large number of snapshots.

The default value of ThreadstackSizekb is 128 and can handle only limited number of snapshots.

If you exceed that number then the host will e disconnected.

Instead of increasing the threadstack value we can delete or consolidate snapshots for virtual machines.

There is a line in the vpxa.cfg file on the host that has an entry <ThreadStackSizeKb> and the value is set to 128.  When vpxa service attempts to connect to vCenter this Thread Stack is limited to not allow VMs that are too many snapshots.


If you don't want the snapshots to be deleted we will increase the threadstacksize so the host can handle more number of snapshtos.


Please let me know if you still have any further queries.


Thanks

Sam





0 Kudos
DanielOprea
Hot Shot
Hot Shot
Jump to solution

Hi,

Please aply this KB:

http://h20564.www2.hp.com/hpsc/doc/public/display?docId=emr_na-c03323800

Drivers & Software - HP Support Center.

Upgrade the rom from the server HP

PLEASE CONSIDER AWARDING any HELPFUL or CORRECT answer. Thanks!!
Por favor CONSIDERA PREMIAR cualquier respuesta ÚTIL o CORRECTA. ¡¡Muchas gracias!!
Blogs: https://danieloprea.blogspot.com/