Summary: | sys-fs/unionfs request for unmask | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | postmodern <brodigan> |
Component: | [OLD] Core system | Assignee: | Stefan Schweizer (RETIRED) <genstef> |
Status: | RESOLVED LATER | ||
Severity: | enhancement | CC: | satya |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
postmodern
2005-07-22 06:16:08 UTC
I disagree: unionfs is perfect for read-only usage, but r/w users can incurr in kernel oops, thread race condition and data losses (particularly in concurrent hard link creations or massive deletions) [1]. The main problems lie in the kernel VFS architecture, which originally was not though for deep abstractions (reiserfs4 is another case in which such limitations are revealed). 1.0.13 fixes some of these (==deletions) limitations, but has issues as well. The development is very promising, though, and I think stabilization could come in a couple of releases. genstef: I think this is a wontfix. [1] http://www.fsl.cs.sunysb.edu/pipermail/unionfs/2005-July/thread.html After posting this request, I found several minor inconsistencies with the unionctl utility along with an occasional Kernel Panic with unioning and deunioning several branches in r/w mode. I agree this is a promising tool but more testing needs to be done. Won't go stable on amd64... maybe it will go stable later, just reopne when you consider it stable. |