Updates to Build Appliance drives it into a bad state

Updates to Build Appliance drives it into a bad state

[Cristian Iorga]

[-- Attachment #1: Type: text/plain, Size: 2171 bytes --]

Hello all,

Because there have been a lot of improvements into Hob after 1.3 Yocto 
Project, I have decided to update the Build Appliance to a more recent 
version of Yocto Linux, and of course, as a consequence, of Hob.

I started with different revisions of the poky tree, gradually setting 
the BA with the poky tree commit 
9e0d3c0faa9cc80ce8f2ce0eb31dcd20ef875955, which is RP's "sstate: Bump 
version number to deal with layout fixes" fix.

However, I see two major issues with this HEAD for YP's BA's Linux tree:

1. The first time the BA starts, it will give an error regarding a 
bitbake timeout related to a command sent by Hob. See 
bitbake-server-command-timeout-2012-11-15 18:14:15.png for details.

Please note that is not happening if running Hob natively on my machine 
with the same repository revision.
Also, the Hob command which causes a timeout is not the same one on 
different runs (the timeout-ed command is different almost on each run).

Now, I managed to track the issue to this commit:

commit c1c20c02a0eb808a099bebfcc7e90188baa22ba4
Author: Christopher Larson <chris_larson@mentor.com>
Date:   Mon Oct 29 13:01:23 2012 -0700

     bitbake: command: add error to return of runCommand

because setting the BA repo version to the previous commit makes the 
error disappear.
Please note that I not saying that this commit is wrong, or anything 
else, because it work correctly on a native Hob.
I am also testing on a modern day machine, but even so the environment 
tests performed by Hob the first time it start in BA take a very long 
time (because of network connectivity tests timeouts).

2. Garbage taken from bitbake (that's my wild guess) in network proxy 
Settings window. See attached screenshot for details 
(garbage_proxy_settings-2012-11-16 20:42:17.png)
The same case here, it is not happening in Hob.
garbage_proxy_settings-2012-11-16 20:42:17.png. Now, if I am correct on 
this one, I would say that if I ran BA in vmWare Player with a single 
core allocated for the BA's VM, this issue is not happening, but I need 
to check to be sure.

Please advise.

Regards,
Cristian Iorga

[-- Attachment #2: bitbake-server-command-timeout-2012-11-15 18:14:15.png --]
[-- Type: image/png, Size: 50046 bytes --]

[-- Attachment #3: garbage_proxy_settings-2012-11-16 20:42:17.png --]
[-- Type: image/png, Size: 86346 bytes --]

Updates to Build Appliance drives it into a bad state

[Cristian Iorga]

[-- Attachment #1: Type: text/plain, Size: 2164 bytes --]



Hello all,

Because there have been a lot of improvements into Hob after 1.3 Yocto
Project, I have decided to update the Build Appliance to a more recent
version of Yocto Linux, and of course, as a consequence, of Hob.

I started with different revisions of the poky tree, gradually setting
the BA with the poky tree commit
9e0d3c0faa9cc80ce8f2ce0eb31dcd20ef875955, which is RP's "sstate: Bump
version number to deal with layout fixes" fix.

However, I see two major issues with this HEAD for YP's BA's Linux tree:

1. The first time the BA starts, it will give an error regarding a
bitbake timeout related to a command sent by Hob. See
bitbake-server-command-timeout-2012-11-15 18:14:15.png for details.

Please note that is not happening if running Hob natively on my machine
with the same repository revision.
Also, the Hob command which causes a timeout is not the same one on
different runs (the timeout-ed command is different almost on each run).

Now, I managed to track the issue to this commit:

commit c1c20c02a0eb808a099bebfcc7e90188baa22ba4
Author: Christopher Larson <chris_larson@mentor.com>
Date:   Mon Oct 29 13:01:23 2012 -0700

      bitbake: command: add error to return of runCommand

because setting the BA repo version to the previous commit makes the
error disappear.
Please note that I not saying that this commit is wrong, or anything
else, because it work correctly on a native Hob.
I am also testing on a modern day machine, but even so the environment
tests performed by Hob the first time it start in BA take a very long
time (because of network connectivity tests timeouts).

2. Garbage taken from bitbake (that's my wild guess) in network proxy
Settings window. See attached screenshot for details
(garbage_proxy_settings-2012-11-16 20:42:17.png)
The same case here, it is not happening in Hob.
garbage_proxy_settings-2012-11-16 20:42:17.png. Now, if I am correct on
this one, I would say that if I ran BA in vmWare Player with a single
core allocated for the BA's VM, this issue is not happening, but I need
to check to be sure.

Please advise.

Regards,
Cristian Iorga




[-- Attachment #2: bitbake-server-command-timeout-2012-11-15 18:14:15.png --]
[-- Type: image/png, Size: 50046 bytes --]

[-- Attachment #3: garbage_proxy_settings-2012-11-16 20:42:17.png --]
[-- Type: image/png, Size: 86346 bytes --]