…
Some time ago I posted about a side-result of the work on the VM migration 3.1⇒3.0 issue. Here is several slides[1] about the pyvfs library[2]. It is not a ready-to-use VDSM application, but some research work. It is not designed specifically for VDSM and can be used in any Python script.
I hope, with the library we can create some features for VDSM, like file-based RPC or internal structures monitoring. Right now I use it to observe VM objects and monitor internal VM config representation.
I would be glad to discuss it and can prepare a short presentation using VDSM as a sample.
[1] slides: http://peet.spb.ru/pyvfs.pdf (53kB) [2] the code on github: https://github.com/svinota/pyvfs
On 10/15/2012 03:52 PM, Peter V. Saveliev wrote:
…
Some time ago I posted about a side-result of the work on the VM migration 3.1⇒3.0 issue. Here is several slides[1] about the pyvfs library[2]. It is not a ready-to-use VDSM application, but some research work. It is not designed specifically for VDSM and can be used in any Python script.
I hope, with the library we can create some features for VDSM, like file-based RPC or internal structures monitoring. Right now I use it to observe VM objects and monitor internal VM config representation.
I would be glad to discuss it and can prepare a short presentation using VDSM as a sample.
[1] slides: http://peet.spb.ru/pyvfs.pdf (53kB) [2] the code on github: https://github.com/svinota/pyvfs
This would be a fantastic debug resource for anyone supporting VDSM (/me waves) but I would be worried about having it mounted rw if at all.
Maybe we could only allow users to mount this via a specific verb or separate log collection tool (sosreport etc). Thoughts?
Lee
vdsm-devel@lists.fedorahosted.org