REISER4 "strange" partition size

REISER4 "strange" partition size

[Paolo Correnti]

Hallo,

as anyone experience with this e-mail object ?

I'm testing kernel 2.6.1 + latest snapshot
and I've noticed that if a use mkreiserfs 
and mkfs.reiser4 on 2 partitions of the same size
with df I obtain that reiser4 partition has "lost"
about 5% of space (for example on a 4 Gb partition
I see 3,8 Gb).

Then making intensive writing test on this reiser4
partition (for example: writing 1 million rows on an
Oracle table in archivelog mode) on each test I obtain
a corrupted redolog file [sometimes is recoverable,
sometimes I've to rebuild the entire DB).


I've never had this kind of problem with kernel 2.6.0
and last 2003 snapshot.

Best regards

Paolo

__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free web site building tool. Try it!
http://webhosting.yahoo.com/ps/sb/

Re: REISER4 "strange" partition size

[Nikita Danilov]

Paolo Correnti writes:
 > Hallo,
 > 
 > as anyone experience with this e-mail object ?
 > 
 > I'm testing kernel 2.6.1 + latest snapshot
 > and I've noticed that if a use mkreiserfs 
 > and mkfs.reiser4 on 2 partitions of the same size
 > with df I obtain that reiser4 partition has "lost"
 > about 5% of space (for example on a 4 Gb partition
 > I see 3,8 Gb).

Good catch!

These 5% are reserved by reiser4 for internal use.

 > 
 > Then making intensive writing test on this reiser4
 > partition (for example: writing 1 million rows on an
 > Oracle table in archivelog mode) on each test I obtain
 > a corrupted redolog file [sometimes is recoverable,
 > sometimes I've to rebuild the entire DB).
 > 

Hmm...

It is not clear how to diagnose this. What oracle are you using? Is it
available freely?

Also you can compile reiser4 with CONFIG_REISER4_CHECK and
CONFIG_REISER4_DEBUG on, but it will run _much_ _more_ _slowly_.

 > 
 > I've never had this kind of problem with kernel 2.6.0
 > and last 2003 snapshot.
 > 
 > Best regards
 > 
 > Paolo
 > 

Nikita.

Re: REISER4 "strange" partition size

[Paolo Correnti]

--- Nikita Danilov <Nikita@Namesys.COM> wrote:

>  > 
>  > Then making intensive writing test on this
> reiser4
>  > partition (for example: writing 1 million rows on
> an
>  > Oracle table in archivelog mode) on each test I
> obtain
>  > a corrupted redolog file [sometimes is
> recoverable,
>  > sometimes I've to rebuild the entire DB).
>  > 
> 
> Hmm...
> 
> It is not clear how to diagnose this. What oracle
> are you using? Is it
> available freely?
> 

I'm using Oracle 9.2 on Linux (Slackware 9.1)

> Also you can compile reiser4 with
> CONFIG_REISER4_CHECK and
> CONFIG_REISER4_DEBUG on, but it will run _much_
> _more_ _slowly_.
> 

OK, I'll try, also if like I've already wrote 
I've never had this kind of problem with kernel
2.6.0 and last 2003 snapshot

Best regards
 
Paolo



__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free web site building tool. Try it!
http://webhosting.yahoo.com/ps/sb/

REISER4 corruptions errors

[Paolo Correnti]

In my last message "REISER4 strange partition size"
I wrote:

"... making intensive writing test on this
reiser4 partition (for example: writing 1 million rows
on an Oracle table in archivelog mode) on each test I
obtain a corrupted redolog file [sometimes is
recoverable, sometimes I've to rebuild the entire
DB)."

And Nikita answered:

"It is not clear how to diagnose this. What oracle
are you using? Is it available freely?"

Really I think this test is not so easy to prepare
(you have to download Oracle, install it and bla bla
...) but I think my last test is more "human".

I recomplied Kernel 2.6.0 with
CONFIG_REISER4_BADBLOCKS=y

I'm using 2.6.0 + 20031223 snapshot because 
2.6.1 + 20040119_fixed always gives me some file
corrupted and I was spending too much time re-creating
Oracle DB.

Reiser4 partition is mounted with tmgr.atom_max_age=20
option.

I noticed that if you copy a large amount of data (at
least 500 Mb) from a partition to a Reiser4 partition,
umounting Reiser4 part and using fsck.reiser4
you always obtain from 5 to 40 (and more) fixable
corruptions detected.

It seems that with 2.6.0 + 20031223 snapshot these
corruptions are not so "dangerous" to corrupt files
(I mean, also with these corruptions I'm able to work
with Oracle). Unfortunately I cannot say the same
with 2.6.1 and latest snapshot.

Any idea ?

All the best

Paolo






__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free web site building tool. Try it!
http://webhosting.yahoo.com/ps/sb/

Re: REISER4 corruptions errors

[Nikita Danilov]

Paolo Correnti writes:
 > In my last message "REISER4 strange partition size"
 > I wrote:
 > 
 > "... making intensive writing test on this
 > reiser4 partition (for example: writing 1 million rows
 > on an Oracle table in archivelog mode) on each test I
 > obtain a corrupted redolog file [sometimes is
 > recoverable, sometimes I've to rebuild the entire
 > DB)."
 > 
 > And Nikita answered:
 > 
 > "It is not clear how to diagnose this. What oracle
 > are you using? Is it available freely?"
 > 
 > Really I think this test is not so easy to prepare
 > (you have to download Oracle, install it and bla bla
 > ...) but I think my last test is more "human".
 > 
 > I recomplied Kernel 2.6.0 with
 > CONFIG_REISER4_BADBLOCKS=y
 > 
 > I'm using 2.6.0 + 20031223 snapshot because 
 > 2.6.1 + 20040119_fixed always gives me some file
 > corrupted and I was spending too much time re-creating
 > Oracle DB.
 > 
 > Reiser4 partition is mounted with tmgr.atom_max_age=20
 > option.
 > 
 > I noticed that if you copy a large amount of data (at
 > least 500 Mb) from a partition to a Reiser4 partition,
 > umounting Reiser4 part and using fsck.reiser4
 > you always obtain from 5 to 40 (and more) fixable
 > corruptions detected.
 > 
 > It seems that with 2.6.0 + 20031223 snapshot these
 > corruptions are not so "dangerous" to corrupt files
 > (I mean, also with these corruptions I'm able to work
 > with Oracle). Unfortunately I cannot say the same
 > with 2.6.1 and latest snapshot.

Well, for all I know, this very well may be a bug in the fsck, rather
than corruption of the on-disk data structures. Are you experiencing any
problems when _using_ this partition (error messages in the kernel log,
crashes, deadlocks, etc.)?

 > 
 > Any idea ?
 > 
 > All the best
 > 
 > Paolo
 >

Nikita.

Re: REISER4 corruptions errors

[Paolo Correnti]

--- Nikita Danilov <Nikita@Namesys.COM> wrote:

> Well, for all I know, this very well may be a bug in
> the fsck, rather than corruption of the on-disk data
> structures. Are
> you experiencing any
> problems when _using_ this partition (error messages
> in the kernel log,
> crashes, deadlocks, etc.)?
> 

I have problems using the "20040119-fixed" partition
with 2.6.1 (after writing many MB of data, almost
always I obtain a file corrupted so that Oracle
doesn't start). With 2.6.0 and 20031223 snapshot I
have no problems using the "20031223" partition.

But in both tests I saw that fsck.reiser4 gave me
from 5 to 35 and more corruptions errors, all of type

Error: Node (210326), item (7): StatData of the file
[10001:1616662635f5445:10002] has the wrong bytes
(3625472), Should be
(3629056). Plugin (stat40).

So I was thinking (perhaps strangely ...) that this
kind of corruption was more dangerous with 2.6.1 +
20040119-fixed (I'm speaking about Oracle logfile
corrupted after 5 million rows written) than with
2.6.0 + 20030123 (which never gave me an Oracle file
corrupted, also after 10 million rows written).

I made the same test on 2 different disks so I'm
almost sure this is not an hardware problem.

Best regards

Paolo


__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free web site building tool. Try it!
http://webhosting.yahoo.com/ps/sb/

Re: REISER4 corruptions errors

[Alex Zarochentsev]

On Mon, Jan 26, 2004 at 06:25:28AM -0800, Paolo Correnti wrote:
> 
> --- Nikita Danilov <Nikita@Namesys.COM> wrote:
> 
> > Well, for all I know, this very well may be a bug in
> > the fsck, rather than corruption of the on-disk data
> > structures. Are
> > you experiencing any
> > problems when _using_ this partition (error messages
> > in the kernel log,
> > crashes, deadlocks, etc.)?
> > 
> 
> I have problems using the "20040119-fixed" partition
> with 2.6.1 (after writing many MB of data, almost
> always I obtain a file corrupted so that Oracle
> doesn't start). With 2.6.0 and 20031223 snapshot I
> have no problems using the "20031223" partition.
> 
> But in both tests I saw that fsck.reiser4 gave me
> from 5 to 35 and more corruptions errors, all of type
> 
> Error: Node (210326), item (7): StatData of the file
> [10001:1616662635f5445:10002] has the wrong bytes
> (3625472), Should be
> (3629056). Plugin (stat40).
> 
> So I was thinking (perhaps strangely ...) that this
> kind of corruption was more dangerous with 2.6.1 +
> 20040119-fixed (I'm speaking about Oracle logfile

I guess it is not a dangerous corruption.  Probaply Oracle does too strong
checks for its log files.  The file content and size should be OK, except
i_blocks and i_bytes fields.  

I think the source of that curruption is in the reiser4 deletion optimization
performed by the cut_tree() routine (inode_sub_bytes() is not called in some
cases).

this patch should help (! not tested):

===== tree.c 1.562 vs edited =====
--- 1.562/tree.c        Wed Jan 14 11:46:20 2004
+++ edited/tree.c       Mon Jan 26 18:24:39 2004
@@ -1468,7 +1468,7 @@
                if ((result != 0) && (result != -E_NO_NEIGHBOR))
                        break;
                /* Check can we delete the node as a whole. */
-               if (iterations && znode_get_level(node) == LEAF_LEVEL &&
+               if (0 && iterations && znode_get_level(node) == LEAF_LEVEL &&
                    UNDER_RW(dk, current_tree, read,
                             keyle(from_key, znode_get_ld_key(node)))) {
                        result = delete_node(next_node_lock.node,


> corrupted after 5 million rows written) than with
> 2.6.0 + 20030123 (which never gave me an Oracle file
> corrupted, also after 10 million rows written).
> 
> I made the same test on 2 different disks so I'm
> almost sure this is not an hardware problem.
> 
> Best regards
> 
> Paolo
> 
> 
> __________________________________
> Do you Yahoo!?
> Yahoo! SiteBuilder - Free web site building tool. Try it!
> http://webhosting.yahoo.com/ps/sb/

-- 
Alex.

Re: REISER4 corruptions errors

[Alex Zarochentsev]

On Mon, Jan 26, 2004 at 05:36:10PM +0100, Carl-Daniel Hailfinger wrote:
> Alex Zarochentsev wrote:
> 
> > this patch should help (! not tested):
> > 
> > ===== tree.c 1.562 vs edited =====
> > --- 1.562/tree.c        Wed Jan 14 11:46:20 2004
> > +++ edited/tree.c       Mon Jan 26 18:24:39 2004
> > @@ -1468,7 +1468,7 @@
> >                 if ((result != 0) && (result != -E_NO_NEIGHBOR))
> >                         break;
> >                 /* Check can we delete the node as a whole. */
> > -               if (iterations && znode_get_level(node) == LEAF_LEVEL &&
> > +               if (0 && iterations && znode_get_level(node) == LEAF_LEVEL &&
> >                     UNDER_RW(dk, current_tree, read,
> >                              keyle(from_key, znode_get_ld_key(node)))) {
> >                         result = delete_node(next_node_lock.node,
> 
> Was it your intention to disable the above code path?

no, it will be fixed in proper way. my fix is a temporary solution.

-- 
Alex.

Re: REISER4 corruptions errors

[Carl-Daniel Hailfinger]

Alex Zarochentsev wrote:

> this patch should help (! not tested):
> 
> ===== tree.c 1.562 vs edited =====
> --- 1.562/tree.c        Wed Jan 14 11:46:20 2004
> +++ edited/tree.c       Mon Jan 26 18:24:39 2004
> @@ -1468,7 +1468,7 @@
>                 if ((result != 0) && (result != -E_NO_NEIGHBOR))
>                         break;
>                 /* Check can we delete the node as a whole. */
> -               if (iterations && znode_get_level(node) == LEAF_LEVEL &&
> +               if (0 && iterations && znode_get_level(node) == LEAF_LEVEL &&
>                     UNDER_RW(dk, current_tree, read,
>                              keyle(from_key, znode_get_ld_key(node)))) {
>                         result = delete_node(next_node_lock.node,

Was it your intention to disable the above code path?


Carl-Daniel
-- 
http://www.hailfinger.org/