Send bitbake cache (from recipe parsing) to another server

Send bitbake cache (from recipe parsing) to another server

[krzysztof.taborski]

[-- Attachment #1: Type: text/plain, Size: 547 bytes --]

Hello,

I am wondering, if it is possible to archive bitbake cache and send to another server.
I tried to archive cache and tmp/cache directories and use them on another server (configs and recipes/workspace name/user name and build server image were same), but for some reason recipes are re-parsed by bitbake.

First of all, I am wondering, if I have even chances to success in such scenario.
If yes I am guessing, that I should modify BB_HASHCONFIG_WHITELIST , but how to check what can be missing?

Best regards,
Krzysztof Taborski

[-- Attachment #2: Type: text/html, Size: 713 bytes --]

Re: [OE-core] Send bitbake cache (from recipe parsing) to another server

[Richard Purdie]

On Fri, 2020-05-22 at 04:59 -0700, krzysztof.taborski@nokia.com wrote:
> I am wondering, if it is possible to archive bitbake cache and send
> to another server.

Only under very specific conditions. The paths and timestamps for the
metadata would all have to match perfectly. The cache is not designed
to be moved anywhere at all.

> I tried to archive cache and tmp/cache directories and use them on
> another server (configs and recipes/workspace name/user name and
> build server image were same), but for some reason recipes are re-
> parsed by bitbake.

Not surprising as there are timestamps stored in them.

> First of all, I am wondering, if I have even chances to success in
> such scenario. 

Low. Its not designed to do that and you should never need to. Why
can't it reparse?

Cheers,

Richard

Re: Send bitbake cache (from recipe parsing) to another server

[krzysztof.taborski]

[-- Attachment #1: Type: text/plain, Size: 400 bytes --]

On Fri, May 22, 2020 at 05:04 AM, Richard Purdie wrote:

> 
> Low. Its not designed to do that and you should never need to. Why
> can't it reparse?

I tried to reduce building time. I know, that it is not a big time, but multiplying by build count it could bring some benefits.
Nevertheless thanks for the answer. At least I know, that I should abandon this idea.

Br,
Krzysztof Taborski

[-- Attachment #2: Type: text/html, Size: 439 bytes --]