From 4be56ff1965828c4d338860f3f1e63fcfa5dd1c4 Mon Sep 17 00:00:00 2001 From: Andrea Mazzoleni Date: Sat, 11 Jan 2014 13:02:34 +0100 Subject: [PATCH] Adds a new TODO entry. --- TODO | 27 ++++++++++++++++++--------- 1 file changed, 18 insertions(+), 9 deletions(-) diff --git a/TODO b/TODO index ec9ef534..ecea3b8a 100644 --- a/TODO +++ b/TODO @@ -5,9 +5,27 @@ This is the list of TODO items for SnapRAID. - Next +* When a silent error is found in sync, try to recover the data, +and if succesfull compute the new parity as if data was read correctly. +But avoid to write the corrected data, and still mark the block as bad. +Keep the present behaviour of not proceeding with sync only if recovering +is not possible. + +* When renaming to .unrecoverable at now we always overwrite +a previous copy losing partial, but potentially better, file recovery. +Is it possible a better strategy ? +Maybe we can rename the unrecoverable as the real file, if the real file +is missing at the first open. This should ensure to always +get better results if the file is hashed. +Something like an incremental recovery, in case multiple fix commands +are attempted. + * Don't disable default 10d/12% if -p or -o is specified. But check that -p0 still work. +* Checks if splitting hash/parity computation in 4K pages +can improve speed in sync. + * Detect files moved from one disk to another. We won't be able to reuse parity, but we can reuse the hash to check that the file was moved correctly. @@ -32,15 +50,6 @@ And create a new block type REP (replaced) where the hash is intented to match the new block and not the previous one like for CHG blocks. -* When renaming to .unrecoverable at now we always overwrite -a previous copy losing partial, but potentially better, file recovery. -Is it possible a better strategy ? -Maybe we can rename the unrecoverable as the real file, if the real file -is missing at the first open. This should ensure to always -get better results if the file is hashed. -Something like an incremental recovery, in case multiple fix commands -are attempted. - - Minor * In the content file save the timestap of the parity files.