on pg_log and neglect ceph journal? It will save lots of
bandwidth, and also based on the consistent pg_log epoch, we can
always recovery data from its peering osd, right? But this will lead
to recovery more objects if the osd crash.
Nicheal
--
To unsubscribe from this list: send the line
Hi Nicheal,
Not only recovery , IMHO the main purpose of ceph journal is to support
transaction semantics since XFS doesn't have that. I guess it can't be achieved
with pg_log/pg_info.
Thanks Regards
Somnath
-Original Message-
From: ceph-devel-ow...@vger.kernel.org
[mailto:ceph-devel
@vger.kernel.org
Subject: RE: puzzled with the design pattern of ceph journal, really ruining
performance
Hi Nicheal,
Not only recovery , IMHO the main purpose of ceph journal is to support
transaction semantics since XFS doesn't have that. I guess it can't be achieved
with pg_log/pg_info
- Mail original -
De: Xiaoxi Chen xiaoxi.c...@intel.com
À: Somnath Roy somnath@sandisk.com, ?? zay11...@gmail.com,
ceph-devel@vger.kernel.org
Envoyé: Mercredi 17 Septembre 2014 09:59:37
Objet: RE: puzzled with the design pattern of ceph journal, really ruining
performance
Hi Nicheal
: Somnath Roy somnath@sandisk.com, ?? zay11...@gmail.com,
ceph-devel@vger.kernel.org
Envoyé: Mercredi 17 Septembre 2014 09:59:37
Objet: RE: puzzled with the design pattern of ceph journal, really ruining
performance
Hi Nicheal,
1. The main purpose of journal is provide transaction semantics (prevent
pattern of ceph journal, really ruining
performance
On 09/17/2014 09:20 AM, Alexandre DERUMIER wrote:
2. Have you got any data to prove the O_DSYNC or fdatasync kill the
performance of journal? In our previous test, the journal SSD (use a
partition of a SSD as a journal for a particular OSD
...@vger.kernel.org
[mailto:ceph-devel-ow...@vger.kernel.org] On Behalf Of Alexandre DERUMIER
Sent: Thursday, September 18, 2014 5:13 AM
To: Mark Nelson
Cc: Somnath Roy; ??; ceph-devel@vger.kernel.org; Chen, Xiaoxi
Subject: Re: puzzled with the design pattern of ceph journal, really ruining
performance
; ??; ceph-devel@vger.kernel.org; Chen, Xiaoxi
Subject: Re: puzzled with the design pattern of ceph journal, really ruining
performance
FWIW, the journal will coalesce writes quickly when there are many
concurrent 4k client writes. Once you hit around 8 4k IOs per OSD,
the journal will start coalescing
don't use aio when closing journal. And
using fsync(). It make the code simple.
How about this?
Thanks!
Jianpeng
ceph-devel@vger.kernel.org
Subject: Re: Deadlock in ceph journal
On 23/08/14 10:22, Somnath Roy wrote:
I think it is using direct io for non-aio mode as well.
Thanks
() on write_thread_entry) and the
result is looks good.
I want to revert the patch which don't use aio when closing journal. And
using fsync(). It make the code simple.
How about this?
Thanks!
Jianpeng
ceph-devel@vger.kernel.org
Subject: Re: Deadlock in ceph journal
On 23/08
);
ceph-devel@vger.kernel.org
Subject: RE: Deadlock in ceph journal
Sounds good. Can you send a patch?
sage
On Mon, 25 Aug 2014, Ma, Jianpeng wrote:
Hi all,
At weekend, I read the kernel code about aio direction. For close(), it
don't wait aio to complete.
But for fsync
On 22/08/14 12:49, Sage Weil wrote:
On Fri, 22 Aug 2014, Mark Kirkwood wrote:
On 22/08/14 03:23, Sage Weil wrote:
I've pushed the patch to wip-filejournal. Mark, can you test please?
I've tested wip-filejournal and looks good (25 test runs, good journal header
each time).
Thanks!
@vger.kernel.org
Subject: Re: Deadlock in ceph journal
On 22/08/14 12:49, Sage Weil wrote:
On Fri, 22 Aug 2014, Mark Kirkwood wrote:
On 22/08/14 03:23, Sage Weil wrote:
I've pushed the patch to wip-filejournal. Mark, can you test please?
I've tested wip-filejournal and looks good (25 test
, August 22, 2014 3:19 PM
To: Sage Weil
Cc: Ma, Jianpeng; Somnath Roy; Samuel Just (sam.j...@inktank.com);
ceph-devel@vger.kernel.org
Subject: Re: Deadlock in ceph journal
On 22/08/14 12:49, Sage Weil wrote:
On Fri, 22 Aug 2014, Mark Kirkwood wrote:
On 22/08/14 03:23, Sage Weil wrote:
I've
On 23/08/14 10:22, Somnath Roy wrote:
I think it is using direct io for non-aio mode as well.
Thanks Regards
Somnath
One thing that does still concern me - if I understand what is happening here
correctly: we write to the journal using aio until we want to stop doing writes
(presumably
Will do.
On 21/08/14 19:30, Ma, Jianpeng wrote:
Mark
After sage merge this into wip-filejournal, can you test again? I think at
present only you can do this work!
--
To unsubscribe from this list: send the line unsubscribe ceph-devel in
the body of a message to majord...@vger.kernel.org
: Deadlock in ceph journal
On Thu, 21 Aug 2014, Ma, Jianpeng wrote:
Yes, Maybe for io_submit, it must use io_getevent. Otherwise the result is
undefined.
If stop_write == true, we don't use aio. How about this way?
That seems reasonable, now that I understand why it doesn't work
On 22/08/14 03:23, Sage Weil wrote:
I've pushed the patch to wip-filejournal. Mark, can you test please?
I've tested wip-filejournal and looks good (25 test runs, good journal
header each time).
Cheers
Mark
--
To unsubscribe from this list: send the line unsubscribe ceph-devel in
the
On Fri, 22 Aug 2014, Mark Kirkwood wrote:
On 22/08/14 03:23, Sage Weil wrote:
I've pushed the patch to wip-filejournal. Mark, can you test please?
I've tested wip-filejournal and looks good (25 test runs, good journal header
each time).
Thanks! Merged.
sage
--
To unsubscribe from
; jianpeng...@intel.com
Subject: RE: Deadlock in ceph journal
On Wed, 20 Aug 2014, Somnath Roy wrote:
Thanks Sage !
So, the latest master should have the fix, right ?
The original patch that caused the regression is reverted, but we'd like to
reapply it if we sort out the issues. wip
: Samuel Just (sam.j...@inktank.com); ceph-devel@vger.kernel.org; Mark
Kirkwood; Ma, Jianpeng
Subject: RE: Deadlock in ceph journal
I suspect what is really needed is a drain_aio() function that will wait for
all
pending aio ops to complete on shutdown. What happens to those IOs if the
process
To: Somnath Roy
Cc: Samuel Just (sam.j...@inktank.com); ceph-devel@vger.kernel.org; Mark
Kirkwood; Ma, Jianpeng
Subject: RE: Deadlock in ceph journal
I suspect what is really needed is a drain_aio() function that will wait
for all
pending aio ops to complete on shutdown. What
: ceph-us...@lists.ceph.com
Subject: Deadlock in ceph journal
Hi Sage/Sam,
During our testing we found a potential deadlock scenario in the filestore
journal code base. This is happening because of two reason.
1. This is because code is not signaling aio_cond from
; jianpeng...@intel.com
Subject: RE: Deadlock in ceph journal
[Copying ceph-devel, dropping ceph-users]
Yeah, that looks like a bug. I pushed wip-filejournal that reapplies
Jianpeng's original patch and this one. I'm not certain about last other
suggested fix, though, but I'm hoping that this fix
...@inktank.com); ceph-devel@vger.kernel.org; Mark
Kirkwood; jianpeng...@intel.com
Subject: RE: Deadlock in ceph journal
Thanks Sage !
So, the latest master should have the fix, right ?
Regards
Somnath
-Original Message-
From: Sage Weil [mailto:sw...@redhat.com]
Sent: Tuesday, August 19, 2014 8:55
@vger.kernel.org; Mark
Kirkwood; jianpeng...@intel.com
Subject: RE: Deadlock in ceph journal
[Copying ceph-devel, dropping ceph-users]
Yeah, that looks like a bug. I pushed wip-filejournal that reapplies
Jianpeng's original patch and this one. I'm not certain about last other
...@intel.com
Subject: RE: Deadlock in ceph journal
On Wed, 20 Aug 2014, Somnath Roy wrote:
Thanks Sage !
So, the latest master should have the fix, right ?
The original patch that caused the regression is reverted, but we'd like to
reapply it if we sort out the issues. wip-filejournal has the offending
Not yet,
If you have to use master either revert commit
4eb18dd487da4cb621dcbecfc475fc0871b356ac or apply the patch for fixing
the hang mentioned here https://github.com/ceph/ceph/pull/2185
Otherwise you could use the wip-filejournal branch which Sage has just
added!
Cheers
Mark
On
Sorry, I see that sage has reverted it.
On 20/08/14 16:58, Mark Kirkwood wrote:
Not yet,
If you have to use master either revert commit
4eb18dd487da4cb621dcbecfc475fc0871b356ac or apply the patch for fixing
the hang mentioned here https://github.com/ceph/ceph/pull/2185
--
To unsubscribe
On Thu, Nov 1, 2012 at 10:33 PM, Gandalf Corvotempesta
gandalf.corvotempe...@gmail.com wrote:
2012/11/1 Mark Nelson mark.nel...@inktank.com:
It will do that for a while, based on how you've tweaked the flush intervals
and various journal settings to determine how much data ceph will allow to
On 11/01/2012 04:18 PM, Gandalf Corvotempesta wrote:
2012/10/31 Stefan Kleijkers ste...@kleijkers.nl:
As far as I know, this is correct. You get a ACK (on the write) back after
it landed on ALL three journals (or/and osds in case of BTRFS in parallel
mode). So If you lose one node, you still
On Wed, Oct 31, 2012 at 2:18 PM, Gandalf Corvotempesta
gandalf.corvotempe...@gmail.com wrote:
In a multi replica cluster (for example, replica = 3) is safe to set
journal on a tmpfs?
As fa as I understood with journal enabled all writes are wrote on
journal and then to disk in a second time.
Hello,
On 10/31/2012 10:24 PM, Tren Blackburn wrote:
On Wed, Oct 31, 2012 at 2:18 PM, Gandalf Corvotempesta
gandalf.corvotempe...@gmail.com wrote:
In a multi replica cluster (for example, replica = 3) is safe to set
journal on a tmpfs?
As fa as I understood with journal enabled all writes are
On Wed, 31 Oct 2012, Tren Blackburn wrote:
On Wed, Oct 31, 2012 at 2:18 PM, Gandalf Corvotempesta
gandalf.corvotempe...@gmail.com wrote:
In a multi replica cluster (for example, replica = 3) is safe to set
journal on a tmpfs?
As fa as I understood with journal enabled all writes are wrote
Hello,
On 10/31/2012 10:58 PM, Gandalf Corvotempesta wrote:
2012/10/31 Tren Blackburn t...@eotnetworks.com:
Unless you're using btrfs which writes to the journal and osd fs
concurrently, if you lose the journal device (such as due to a
reboot), you've lost the osd device, requiring it to be
Hi,
Personally I won't take the risk to loose transactions. If a client
writes into a journal, assuming it's the first write and if the server
crashs for whatever reason, you have high risk of un-consistent data.
Because you just lost what was in the journal.
Tmpfs is the cheapest solution for
36 matches
Mail list logo