Arch Linux

Gullible Jones

Member

Registered: 2004-12-29

Posts: 4,863

Fixing stupidly long ReiserFS mount times

Here's something interesting - the reason for ReiserFS's long mount times. Apparently there is some caching going on on mount, slowing things down; there's a patch to get rid of this (endorsed by Hans Reiser, though he was at first opposed to it), but it seems to be a -mm kernel thing. Perhaps the patch ought to be included in the stock kernel, or at least in -beyond? Or do you guys think that such minor patching would be a bad idea?

(You know, it would have been nice if this had been implemented as a mount option... "nobcache", perhaps, or maybe have "bcache" and make no on-mount bitmap caching the default.)

iphitus

Forum Fellow

From: Melbourne, Australia

Registered: 2004-10-09

Posts: 4,927

Re: Fixing stupidly long ReiserFS mount times

File a bug if you want it in our kernels with a link to some patches for the current kernel, and any relevant, preferably recent discussion on LKML.

A description of what you're speaking of might help too... 'long mount times' means nothing.

Alternatively, fix the actual problem and just use a better filesystem.

James

nightfrost

Member

From: Sweden

Registered: 2005-04-16

Posts: 647

Re: Fixing stupidly long ReiserFS mount times

I actually stopped using reiserfs because of the mount times. Good to know there's some sort of a solution for it. Thanks for the link.

eWoud

Member

Registered: 2005-07-06

Posts: 39

Re: Fixing stupidly long ReiserFS mount times

File a bug if you want it in our kernels with a link to some patches for the current kernel, and any relevant, preferably recent discussion on LKML.

A description of what you're speaking of might help too... 'long mount times' means nothing.

Actually, it's the perfect description of the only real downside of Reiser36, mount times that go up with partition size, in a linear way. Mounting a 150GB partition on a modern hard disk takes about 7seconds.
I tried the beta version of Jeff's patch sometime ago, I'm glad there's a stable version available now. I'm not sure if patch no. 4 is usable without the other 3, but it's worth keeping an eye on.

Alternatively, fix the actual problem and just use a better filesystem.

Let's not go into another "which filesystem is best thread". Ext3+dir_index and Reiser36 are on the same level, speed-wise. Only Reiser36 uses less space (yes, that's with tail packing disabled).

---

tea is overrated

baze

Member

Registered: 2005-10-30

Posts: 393

Re: Fixing stupidly long ReiserFS mount times

btw, 2.6.19 should fix those long mount times for reiserfs if i understand this correct:

REISER: On-demand bitmap loading - speeds up mounting speed (commit) and eliminates minimum window size for bitmap searching - improves the allocator in some corner-cases (commit)