All of lore.kernel.org
 help / color / mirror / Atom feed
* How to prevent the user repeat-login to avoid the ikvm overload
@ 2019-06-28  1:53 xiuzhi
  2019-07-08 20:08 ` Joseph Reynolds
  2020-04-30  3:18 ` ikvm suspend on power cycle test xiuzhi
  0 siblings, 2 replies; 3+ messages in thread
From: xiuzhi @ 2019-06-28  1:53 UTC (permalink / raw)
  To: jae.hyun.yoo, openbmc, eajames, Joel Stanley, Andrew.Jeffery, ratagupt

[-- Attachment #1: Type: text/plain, Size: 977 bytes --]

Hi Jae,Ratan,
   The ikvm or openbmc phosphor-rest process will deadlock when users operate the ikvm on different PC at the same time . The user can't login when the  CPU usage of phosphor-gevent process more than 70%.  This issue  will be triggered when one user login  both on different IP and operate the host by  ikvm.
  My test case:
1, The user "root" login throw AST2500  network  card, enter the ikvm webpage ,the IP 192.168.0.100.
2, The user "root" login throw OCP network card with a different IP 192.168.1.100, enter the ikvm web page, do some keyboard and mouse operations
3, The openbmc will deadlock and cann't login, I must restart the phosphor-gevent service ,then I can login again.
  I created other common users, it is impossible to forbid them to login and operate the ikvm at the same time to cause the bmc overload .
   My personal idea is to allow  one user to login only on one machine at a time. Do you have any good ideas?


  Best,
Xiuzhi

[-- Attachment #2: Type: text/html, Size: 1728 bytes --]

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: How to prevent the user repeat-login to avoid the ikvm overload
  2019-06-28  1:53 How to prevent the user repeat-login to avoid the ikvm overload xiuzhi
@ 2019-07-08 20:08 ` Joseph Reynolds
  2020-04-30  3:18 ` ikvm suspend on power cycle test xiuzhi
  1 sibling, 0 replies; 3+ messages in thread
From: Joseph Reynolds @ 2019-07-08 20:08 UTC (permalink / raw)
  To: xiuzhi, jae.hyun.yoo, openbmc, eajames, Joel Stanley,
	Andrew.Jeffery, ratagupt


On 6/27/19 8:53 PM, xiuzhi wrote:
> Hi Jae,Ratan,
>    The ikvm or openbmc phosphor-rest process will deadlock when users 
> operate the ikvm on different PC at the same time . The user can't 
> login when the CPU usage of phosphor-gevent process more than 70%.  
> This issue  will be triggered when one user login  both on different 
> IP and operate the host by  ikvm.
>   My test case:
> 1, The user "root" login throw AST2500  network  card, enter the ikvm 
> webpage ,the IP 192.168.0.100.
> 2, The user "root" login throw OCP network card with a different IP 
> 192.168.1.100, enter the ikvm web page, do some keyboard and mouse 
> operations
> 3, The openbmc will deadlock and cann't login, I must restart the 
> phosphor-gevent service ,then I can login again.
>   I created other common users, it is impossible to forbid them to 
> login and operate the ikvm at the same time to cause the bmc overload .
>    My personal idea is to allow  one user to login only on one machine 
> at a time. Do you have any good ideas?
I haven't seen a reply to this.  Here is my 2 cents worth:

Enhance the service to establish a mutex (or exclusive lock) scoped to 
the BMC when accessing the ikvm service, and release it when the ikvm 
session ends.  A second attempt to use the service concurrently will 
fail to obtain the lock and if possible, give a nice error message.
Can you determine when the ikvm session ends?
Have an easy way so you can (a) to determine if the lock is held, and 
(b) to clear the lock.  I've seen lock files used for this purpose.  For 
example, flock /tmp/ikvm.lock.  Rebooting the BMC should erase the /tmp 
file which will also clear the lock.

Would that help solve the problem?
- Joseph
>
>   Best,
> Xiuzhi

^ permalink raw reply	[flat|nested] 3+ messages in thread

* ikvm suspend on power cycle test
  2019-06-28  1:53 How to prevent the user repeat-login to avoid the ikvm overload xiuzhi
  2019-07-08 20:08 ` Joseph Reynolds
@ 2020-04-30  3:18 ` xiuzhi
  1 sibling, 0 replies; 3+ messages in thread
From: xiuzhi @ 2020-04-30  3:18 UTC (permalink / raw)
  To: jae.hyun.yoo, openbmc, eajames, Joel Stanley, Andrew.Jeffery

[-- Attachment #1: Type: text/plain, Size: 871 bytes --]

Hi Jae,Eajames,
   The ipkvm will suspend on power cycling test. The error often occurs on  happen on ast2400 ,and  the probability of ast2500 is relatively small.

Test case: ipmitool power reset


The host screen error  read:
usb 1-5.1: device descriptor read/64,error -85
usb 1-5.1: device descriptor read/64,error -85
usb 1-5.1: device descriptor read/64,error -85
usb 1-5.1: device descriptor read/64,error -85
usb 1-5.1: device not accepting address 6,error -85
usb 1-5.1: device not accepting address 7,error -85
usb 1-5-port1:unable to enumerate USB devices




bmc console:
#cat /sys/bus/platform/devices/1e6a0000.usb-vhub/1e6a0000.usb-vhub\:p1/gadget/suspend 

#1
The ipkvm can't work anymore by restarting the start-ipkvm.service.
What causes this problem,bmc kvm, bios or host OS kernel ? 

Any ideas to solve it?



Best,
Xiuzhi

[-- Attachment #2: Type: text/html, Size: 1190 bytes --]

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2020-04-30  3:31 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-06-28  1:53 How to prevent the user repeat-login to avoid the ikvm overload xiuzhi
2019-07-08 20:08 ` Joseph Reynolds
2020-04-30  3:18 ` ikvm suspend on power cycle test xiuzhi

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.