On Mon, Aug 13, 2012 at 6:19 PM, Jeff Janes wrote:
> On Sat, Jun 9, 2012 at 5:43 AM, Tom Lane wrote:
>> Simon Riggs writes:
>>> So now the standard for my patches is that I must consider what will
>>> happen if the xlog is deleted?
>>
>> When you're messing around with something that affects dat
Bruce Momjian writes:
> Stupid question, but why are we not just setting a boolean variable in
> shared memory if we WAL-write a non-XLOG_RUNNING_XACTS record, and only
> checkpoint if that is true?
Well, (1) we are trying to avoid adding such logic to the critical
section inside XLogInsert, and
On Wed, Jun 06, 2012 at 06:46:37PM -0400, Tom Lane wrote:
> I wrote:
> > Actually, it looks like there is an extremely simple way to handle this,
> > which is to move the call of LogStandbySnapshot (which generates the WAL
> > record in question) to before the checkpoint's REDO pointer is set, but
Simon Riggs writes:
> So now the standard for my patches is that I must consider what will
> happen if the xlog is deleted?
When you're messing around with something that affects data integrity, yes.
The long and the short of it is that this patch does reduce our ability
to recover from easily-fo
On Fri, Jun 8, 2012 at 12:24 PM, Kevin Grittner
wrote:
> I haven't been exactly clear on the risks about which Tom and Robert
> have been concerned; is it a question about whether we change the
> meaning of these settings to something more complicated?:
>
> checkpoint_segments (integer)
> Maxim
Robert Haas wrote:
> if the database has checkpointed
I haven't been exactly clear on the risks about which Tom and Robert
have been concerned; is it a question about whether we change the
meaning of these settings to something more complicated?:
checkpoint_segments (integer)
Maximum num
On Thu, Jun 7, 2012 at 9:25 PM, Simon Riggs wrote:
> The only risk of data loss is in the case where someone deletes their
> pg_xlog and who didn't take a backup in all that time, which is hardly
> recommended behaviour. We're at exactly the same risk of data loss if
> someone deletes their pg_clo
On 8 June 2012 05:01, Tom Lane wrote:
> Peter Geoghegan writes:
>> On 7 June 2012 18:03, Robert Haas wrote:
>>> On Thu, Jun 7, 2012 at 12:52 PM, Simon Riggs wrote:
Clearly, delaying checkpoint indefinitely would be a high risk choice.
But they won't be delayed indefinitely, since chan
Peter Geoghegan writes:
> On 7 June 2012 18:03, Robert Haas wrote:
>> On Thu, Jun 7, 2012 at 12:52 PM, Simon Riggs wrote:
>>> Clearly, delaying checkpoint indefinitely would be a high risk choice.
>>> But they won't be delayed indefinitely, since changes cause WAL
>>> records to be written and t
On 7 June 2012 18:03, Robert Haas wrote:
> On Thu, Jun 7, 2012 at 12:52 PM, Simon Riggs wrote:
>> Clearly, delaying checkpoint indefinitely would be a high risk choice.
>> But they won't be delayed indefinitely, since changes cause WAL
>> records to be written and that would soon cause another ch
On 7 June 2012 18:03, Robert Haas wrote:
> On Thu, Jun 7, 2012 at 12:52 PM, Simon Riggs wrote:
>> Clearly, delaying checkpoint indefinitely would be a high risk choice.
>> But they won't be delayed indefinitely, since changes cause WAL
>> records to be written and that would soon cause another ch
Robert Haas writes:
> ... It's better to have a few unnecessary
> checkpoints than to risk losing somebody's data, especially since the
> unnecessary checkpoints only happen with wal_level=hot_standby, but
> the data loss risk exists for everyone.
Yeah, that's another point here: the benefit of t
Simon Riggs writes:
> On 7 June 2012 17:27, Tom Lane wrote:
>> Simon Riggs writes:
>>> If that is the concern, then its a one line fix to add the missing clog
>>> flush.
>> To where, and what performance impact will that have?
> To the point where we decide to skip the other parts of the
> ch
On Thu, Jun 7, 2012 at 12:52 PM, Simon Riggs wrote:
> Clearly, delaying checkpoint indefinitely would be a high risk choice.
> But they won't be delayed indefinitely, since changes cause WAL
> records to be written and that would soon cause another checkpoint.
But that's exactly the problem - it
On 7 June 2012 17:27, Tom Lane wrote:
> Simon Riggs writes:
>>> Robert Haas wrote:
I know of real customers who would have suffered real data loss
had this code been present in the server version they were using.
>
>> If that is the concern, then its a one line fix to add the missing c
Simon Riggs writes:
>> Robert Haas wrote:
>>> I know of real customers who would have suffered real data loss
>>> had this code been present in the server version they were using.
> If that is the concern, then its a one line fix to add the missing clog flush.
To where, and what performance imp
On 7 June 2012 14:59, Kevin Grittner wrote:
> Tom Lane wrote:
>
>> there is no guarantee that we'll manage to reach a database state
>> that is consistent with data already flushed out to disk during
>> the last checkpoint.
>
> Robert Haas wrote:
>
>> I know of real customers who would have suff
Robert Haas writes:
> On Wed, Jun 6, 2012 at 6:46 PM, Tom Lane wrote:
>> If we don't like that, I can think of a couple of other ways to get there,
>> but they have their own downsides:
>>
>> * Instead of trying to detect after-the-fact whether any concurrent
>> WAL activity happened during the
On Wed, Jun 6, 2012 at 6:46 PM, Tom Lane wrote:
> I wrote:
>> Actually, it looks like there is an extremely simple way to handle this,
>> which is to move the call of LogStandbySnapshot (which generates the WAL
>> record in question) to before the checkpoint's REDO pointer is set, but
>> after we
Tom Lane wrote:
> there is no guarantee that we'll manage to reach a database state
> that is consistent with data already flushed out to disk during
> the last checkpoint.
Robert Haas wrote:
> I know of real customers who would have suffered real data loss
> had this code been present in t
On 6 June 2012 20:08, Tom Lane wrote:
> In commit 18fb9d8d21a28caddb72c7ffbdd7b96d52ff9724, Simon modified the
> rule for when to skip checkpoints on the grounds that not enough
> activity has happened since the last one. However, that commit left the
> comment block about it in a nonsensical st
I wrote:
> Actually, it looks like there is an extremely simple way to handle this,
> which is to move the call of LogStandbySnapshot (which generates the WAL
> record in question) to before the checkpoint's REDO pointer is set, but
> after we have decided that we need a checkpoint.
On further con
Robert Haas writes:
> On Wed, Jun 6, 2012 at 4:24 PM, Tom Lane wrote:
>>> I felt (and still feel) that this was misguided.
>> Looking at it again, I'm inclined to agree. The behavior was entirely
>> correct up until somebody decided to emit a continuing stream of
>> XLOG_RUNNING_XACTS WAL recor
On Wed, Jun 6, 2012 at 4:24 PM, Tom Lane wrote:
>> I felt (and still feel) that this was misguided.
>
> Looking at it again, I'm inclined to agree. The behavior was entirely
> correct up until somebody decided to emit a continuing stream of
> XLOG_RUNNING_XACTS WAL records even when the system is
Robert Haas writes:
> On Wed, Jun 6, 2012 at 3:08 PM, Tom Lane wrote:
>> In commit 18fb9d8d21a28caddb72c7ffbdd7b96d52ff9724, Simon modified the
>> rule for when to skip checkpoints on the grounds that not enough
>> activity has happened since the last one.
> IIRC, the inspiration for the change
On Wed, Jun 6, 2012 at 3:08 PM, Tom Lane wrote:
> In commit 18fb9d8d21a28caddb72c7ffbdd7b96d52ff9724, Simon modified the
> rule for when to skip checkpoints on the grounds that not enough
> activity has happened since the last one. However, that commit left the
> comment block about it in a nonse
In commit 18fb9d8d21a28caddb72c7ffbdd7b96d52ff9724, Simon modified the
rule for when to skip checkpoints on the grounds that not enough
activity has happened since the last one. However, that commit left the
comment block about it in a nonsensical state:
* If this isn't a shutdown or forced c
27 matches
Mail list logo