2,257
edits
Changes
Jump to navigation
Jump to search
Created page with "This is the plan for image cache and proxy further development before it gets into master == Document the whole thing == Add articles for newly introduced actions and option..."
This is the plan for image cache and proxy further development before it gets into master
== Document the whole thing ==
Add articles for newly introduced actions and options to the [[CLI]] page. Also, it would be good to have an article describing the protocols involved.
== Make the unix sockets reside in work-dir ==
Looking at do_open_remote_image() I see that it fchdir-s to image dir before connecting to proxy/cache. Better solution is to put the socket into [[directories|workdir]].
After this the option {{Opt|--images-dir}} should become optional. Provided the {{Opt|--remote}} is given CRIU should work purely on the work-dir and not generate anything in the [[directories|images-dir]].
== Tune up the image_cache and image_proxy commands to obey daemon CLI rules ==
They should accept the {{Opt|--status-fd}} and {{Opt|--pidfile}} options. Presumably the very <code>cr_daemon()</code> call should be equipped with everything that should be done for [[daemonizing|CLI/opt/--daemon]] and proxy/cache tasks should just call it :)
== Fix local connections not to generate per-image threads ==
There can be many images and it's not nice to stress the system with such amount of threads. The <code>criu/uffd.c</code> manages multiple descriptors with page-faults using the epoll stuff and can be used as an example.
The <code>accept_remote_image_connections()</code> seem not to work well with <code>opts.ps_socket</code> [[scenario|RPC]] as the former just calls <code>accept()</code> on whatever socket is passed there, while the <code>opts.ps_socket</code> is already an established socket for data transfer.
== No strings in protocol ==
Now the hard-coded "RESTORE_FINISH" string (and DUMP_FINISHED one) is used to terminate the communication. Need to tune up the protobuf objects to send boolean (or integer) EOF sign rather that the string.
== Check how proxy/cache works with incremental dumps ==
Looking at the <code>skip_remote_bytes()</code> I think that image-cache and -proxy still do not work well with stacked pages images. Probably for those we'll need the [[page server]] or [[userfaultfd|lazy pages]] like protocol that would request the needed regions and receive it back rather than read bytes from sockets simply to skip those.
== Add support for cache/proxy into go-phaul code ==
I haven't yet finished with the prototype, but plan to do it soon, so once the above steps are done we'll be able to proceed with this one.
[[Category: Development]]
[[Category: Plans]]
== Document the whole thing ==
Add articles for newly introduced actions and options to the [[CLI]] page. Also, it would be good to have an article describing the protocols involved.
== Make the unix sockets reside in work-dir ==
Looking at do_open_remote_image() I see that it fchdir-s to image dir before connecting to proxy/cache. Better solution is to put the socket into [[directories|workdir]].
After this the option {{Opt|--images-dir}} should become optional. Provided the {{Opt|--remote}} is given CRIU should work purely on the work-dir and not generate anything in the [[directories|images-dir]].
== Tune up the image_cache and image_proxy commands to obey daemon CLI rules ==
They should accept the {{Opt|--status-fd}} and {{Opt|--pidfile}} options. Presumably the very <code>cr_daemon()</code> call should be equipped with everything that should be done for [[daemonizing|CLI/opt/--daemon]] and proxy/cache tasks should just call it :)
== Fix local connections not to generate per-image threads ==
There can be many images and it's not nice to stress the system with such amount of threads. The <code>criu/uffd.c</code> manages multiple descriptors with page-faults using the epoll stuff and can be used as an example.
The <code>accept_remote_image_connections()</code> seem not to work well with <code>opts.ps_socket</code> [[scenario|RPC]] as the former just calls <code>accept()</code> on whatever socket is passed there, while the <code>opts.ps_socket</code> is already an established socket for data transfer.
== No strings in protocol ==
Now the hard-coded "RESTORE_FINISH" string (and DUMP_FINISHED one) is used to terminate the communication. Need to tune up the protobuf objects to send boolean (or integer) EOF sign rather that the string.
== Check how proxy/cache works with incremental dumps ==
Looking at the <code>skip_remote_bytes()</code> I think that image-cache and -proxy still do not work well with stacked pages images. Probably for those we'll need the [[page server]] or [[userfaultfd|lazy pages]] like protocol that would request the needed regions and receive it back rather than read bytes from sockets simply to skip those.
== Add support for cache/proxy into go-phaul code ==
I haven't yet finished with the prototype, but plan to do it soon, so once the above steps are done we'll be able to proceed with this one.
[[Category: Development]]
[[Category: Plans]]