search for: braam

---------- Forwarded Message ---------- Subject: ext3 for 2.4 Date: Thu, 17 May 2001 21:20:38 +1000 From: Andrew Morton <andrewm@uow.edu.au> To: ext2-devel@lists.sourceforge.net, "Peter J. Braam" <braam@mountainviewdata.com>, Andreas Dilger <adilger@turbolinux.com>, "Stephen C. Tweedie" <sct@redhat.com> Cc: linux-fsdevel@vger.kernel.org Summary: ext3 works, page_launder() doesn't :) The tree is based on the porting work which Peter Braam did. It...

Hi, We hit an interesting issue with Ext3 in ordered data mode. A large file was truncated (open with O_TRUNC) and immediately over-written with a single write call. The write call returned ENOSPACE, but a check for space (using essentially statfs) promised that there would be plenty of space. The reason turned out to be that the blocks freed by the truncation were not yet available to the

Hi, Arthur found out that ext3 is not changing the ctime on the "old_dir" (the object that is being renamed), but ext2 does. It looks to me like this is simply an omission of the following little patch from namei.c - Peter - --- fs/ext3/namei.c.orig Mon Mar 19 22:55:03 2001 +++ fs/ext3/namei.c Mon Mar 19 22:53:40 2001 @@ -985,6 +985,13 @@ new_dir->i_version =

...; > Versions: > 2.2.19 + 0.0.7a > 2.4.5 + 0.0.6 > > PLEASE USE THESE AT YOUR OWN RISK - THEY CONTAIN EXPERIMENTAL FILE SYSTEM > CODE. > > The RPMS's can be found at: > > ftp://ftp.clusterfilesystem.com/pub/ext3 > > Have a good day! > > - Peter J. Braam - > http://www.clusterfilesystem.com -- Levente "The only thing worse than not knowing the truth is ruining the bliss of ignorance." "Si vis pacem para bellum!"

Hi, I have a detailed question about ordered data writes. Suppose we (i.e. intermezzo) does a transaction, which is closed and then followed by an ordered write. Let's assume the ordered write doesn't require new allocation metadata. Is the write still postponed until the first transaction has committed? As far as I can see, prepare_write always starts a transaction, but I'm

Hi, I noticed that using a sparse file as a loop device leads to non journaled block allocation in Ext3. Is this something that should be fixed? - Peter -

Hi Stephen, We had some starvation/locks happening to us under very heavy load in two cases: - InterMezzo asked ext3 to do a journaled file write (for 1 block) essentially using ext3_write - similarly for truncate. These lockups went away when we started the transaction in InterMezzo and reserved somewhat more space than ext3 does. Any clues as to what this might be? Are the ext3

Hi, I'm running 0.03b and it appears that after a clean unmount the needs_recovery flag is still set. [This makes is quite cumbersome to switch between ext2 and ext3.] How can this be fixed? Oh, and I'm very happy. - Peter -

...rk, (was not log on to the samba server) with a fixed IP, xxx.xxx.xxx.123 no packets are dropped. I do not think it is a problem with the SuSE DHCP as it occurs with a fixed IP windows pc also. Unless there is some setting I have to make on the windows machines. Hope you can help. Regards Braam

I notice that raid6 is recommended in the manual. eg. "...RAID6 is a must." http://manual.lustre.org/manual/LustreManual16_HTML/DynamicHTML-11-1.html which I found a bit surprising given that in Dec ''06 Peter Braam said on this list "Some of our customers experienced data corruption in the RAID6 layer.", and recommending the CFS optimised md raid5 path through the kernel. I can''t see any patches in lustre 1.6.3 that apply to a raid6 path, although perhaps none of these raid* patches are ap...

Hello, I have today released an initial version of extended attributes and access control lists for ext3 (patch against the 2.4.13-ac3 kernel). Eric Jarman <ejarman@acm.org> has contributed a lot to this effort. Since I'm not very much into the innards of ext3, can some of you please take a look at the patch, and see whether it contains any flaws (and tell me which flaws)? Thanks! The

...of zeroes in the middle. This seems somewhat suspicious, but it's likely that I'm not understanding everything about ext3. Thanks for clarifying the situation. - Peter - -----Original Message----- From: Aaron Smith [mailto:aaron@fry.dyn.us.mvd] Sent: Monday, March 12, 2001 5:50 PM To: braam@mountainviewdata.com Subject: /* * Intermezzo. (C) 1998 Peter J. Braam */ #include <linux/types.h> #include <linux/kernel.h> #include <linux/fs.h> #include <linux/malloc.h> #include <linux/vmalloc.h> #include <linux/time.h> #include <linux/errno.h> #in...

CFS has openings for Systems Software Engineers and Principal Systems Software Eningeers. For details see http://www.clusterfs.com/jobs.html Join our fun team! - Peter -

Hi, We have encountered another funny with ext3 on 2.4. Like the kernel NFS server, we have a routine in InterMezzo that does something like looking up an inode by inode number. (compare intermezzo's presto_ilookup or knfsd's nfsfh_iget) Effectively both of these routines do ilookup() { inode = iget(sb, ino) if (inode->i_nlink==0) iput(inode); .... } We find that

Hi Alan, Would you be opposed to including an InterMezzo patch for the ac series soon? It doesn't touch anything outside of its fs directory. It would probably help us get a few more users and fits nicely with ext3 being in ac now. We are also quite far along Reiser support for InterMezzo. Please let me know your thoughts on this. Thanks! - Peter -

Hi, Yesterday my ext3 / on 2.4.5 recovered (after suddenly powering off the machine) and messed up an X11 lock file in /tmp. I inspected the file system with debugfs and found that inode was marked deleted, but the directory entry had not been removed. Weird. - Peter -

Hi, I built 2.4.5 kernel rpms with the 0.0.8 ext3 code. The Linux base is 2.4.5-ac17. I have been running this happily. Please use this at your own risk - it contains experimental file system code! You can find the packages at: ftp://ftp.clusterfs.com/pub/ext3 Enjoy! - Peter -

Hi Andreas, Stephen, We have a lot working now: 1. journal recovery and initialization stuff is working 2. transactions go to the disk 3. infrastructure is there to do transcactions 4. ext3_create is fully operational. The problems we have seen mostly have to do with differences in which buffer heads are being initialized. Things like b_transaction etc. were not cleaned up. There are more

>> 1 Use case table. >> >> i. we decided on an implementation constraint to store both file layouts >> as attributes of objects in a redirection layer in the client. These >> layouts would be obtained from the MDS cluster, see (iv) >> >> ii. We decided that locks would be taken in a hierarchical manner - >> the flash cache would run a LDLM and

Hi Nikita - This looks excellent, except that I don''t feel we have a good basis for the estimates yet. This has major architectural value as it gives a component breakdown and should be recorded as such on the architecture wiki. When you do these component breakdowns it is important to identify what interfaces are offered and used by the components (this is the static aspect of