Re: [freenet-support] Datastore resize

2012-10-10 Thread Jep

Matthew Toseland:

On Friday 05 Oct 2012 15:00:13 Jep wrote:

Matthew Toseland:

On Thursday 04 Oct 2012 21:12:36 Jep wrote:
Datastore(PUBKEY-cache) maintenance in progress: xxx/xxx. We have 
changed the datastore so that it will cause less load on your computer 
(particularly disk accesses), but we have to rebuild the indexes first..


This is coming up every time I have to reboot.
Who or what is 'we', is this planned, and if so, why haven't I anywhere 
seen it announced that this is supposed to happen?

Because it was implemented months ago?

And is it intended we do this each and every bootup?

Also months ago I haven't seen anything on it.


No, clearly not. It's supposed to switch to the slot filters and then live 
happily ever after. Hmmm ...

Apparently we use the same message when migrating to slot filters from bloom 
filters as we do when rebuilding because of an unclean shutdown. I have fixed 
this in the source, it will be included in the next build; we will use a 
different message if the .slotfilter file doesn't exist.

In your datastore/ folder you should have the following .slotfilter files, as 
well as other stuff:

CHK-cache.slotfilter
CHK-clientcache.slotfilter
CHK-store.slotfilter
PUBKEY-cache.slotfilter
PUBKEY-clientcache.slotfilter
PUBKEY-store.slotfilter
SSK-cache.slotfilter
SSK-clientcache.slotfilter
SSK-store.slotfilter

Do all these files exist? If so then it probably isn't migrating - it's 
probably rebuilding because of an unclean shutdown. In other words, Freenet has 
been killed by the wrapper because of taking too long to shut down, or you had 
a power cut, or pushed the power button or something.

wrapper.log may reveal whether Freenet is regularly having problems shutting 
down?


Those files are present allright, do have system freezes so that's the 
problem.

Thanks for the response!
___
Support mailing list
Support@freenetproject.org
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe


Re: [freenet-support] Datastore resize

2012-10-09 Thread Matthew Toseland
On Friday 05 Oct 2012 15:00:13 Jep wrote:
 Matthew Toseland:
  On Thursday 04 Oct 2012 21:12:36 Jep wrote:
  Datastore(PUBKEY-cache) maintenance in progress: xxx/xxx. We have 
  changed the datastore so that it will cause less load on your computer 
  (particularly disk accesses), but we have to rebuild the indexes first..
 
  This is coming up every time I have to reboot.
  Who or what is 'we', is this planned, and if so, why haven't I anywhere 
  seen it announced that this is supposed to happen?
  
  Because it was implemented months ago?
 
 And is it intended we do this each and every bootup?
 
 Also months ago I haven't seen anything on it.

No, clearly not. It's supposed to switch to the slot filters and then live 
happily ever after. Hmmm ...

Apparently we use the same message when migrating to slot filters from bloom 
filters as we do when rebuilding because of an unclean shutdown. I have fixed 
this in the source, it will be included in the next build; we will use a 
different message if the .slotfilter file doesn't exist.

In your datastore/ folder you should have the following .slotfilter files, as 
well as other stuff:

CHK-cache.slotfilter
CHK-clientcache.slotfilter
CHK-store.slotfilter
PUBKEY-cache.slotfilter
PUBKEY-clientcache.slotfilter
PUBKEY-store.slotfilter
SSK-cache.slotfilter
SSK-clientcache.slotfilter
SSK-store.slotfilter

Do all these files exist? If so then it probably isn't migrating - it's 
probably rebuilding because of an unclean shutdown. In other words, Freenet has 
been killed by the wrapper because of taking too long to shut down, or you had 
a power cut, or pushed the power button or something.

wrapper.log may reveal whether Freenet is regularly having problems shutting 
down?


signature.asc
Description: This is a digitally signed message part.
___
Support mailing list
Support@freenetproject.org
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe

Re: [freenet-support] Datastore resize

2012-10-04 Thread Matthew Toseland
On Friday 28 Sep 2012 21:54:45 Pascal wrote:
 With the 1st stage of the rebuild taking 3 weeks, and 1416 becoming 
 mandatory in 2 weeks, if I had just waited my node would be unusable for 
 a week.  I noticed an option to do the resize at startup instead that 
 was supposed to be faster, so I set that and upgraded.  I let it run for 
 a bit and then did the math, it was still going to take 3 weeks.  CPU 
 and HDD were both practically idle.  I don't know what a resize does, 
 but it sure seems like there is a bug in there somewhere.

It sleeps 100ms after processing a batch of 128 blocks, to try to avoid severe 
performance effects on the rest of the system. Resize on startup turns this 
off, so *should* be slightly faster. I have no idea why it was so slow for you 
but I imagine it has to do with hard disk seeks being slow - the number of 
bytes transferred could well be relatively low even though the hard disk is 
seeking constantly.
 
 -Pascal
 
 
 On 9/28/2012 4:10 AM, Matthew Toseland wrote:
  On Tuesday 25 Sep 2012 02:38:07 Pascal wrote:
  Changed my datastore size from 750g to 850g just over 5 hours ago.
  It now shows Datastore(CHK-store) resize in progress: 108040/10717960.
 At this rate it will take about 3 weeks just to finish CHK-store.
  Does it them take another 3 weeks to do CHK-cache?  (They were the same
  size beforehand).
 
  Datastore resizing is slow. Especially if it's already mostly full. :(
 ___
 Support mailing list
 Support@freenetproject.org
 http://news.gmane.org/gmane.network.freenet.support
 Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
 Or mailto:support-requ...@freenetproject.org?subject=unsubscribe
 
 


signature.asc
Description: This is a digitally signed message part.
___
Support mailing list
Support@freenetproject.org
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe

Re: [freenet-support] Datastore resize

2012-09-28 Thread Matthew Toseland
On Tuesday 25 Sep 2012 02:38:07 Pascal wrote:
 Changed my datastore size from 750g to 850g just over 5 hours ago. 
 It now shows Datastore(CHK-store) resize in progress: 108040/10717960. 
   At this rate it will take about 3 weeks just to finish CHK-store. 
 Does it them take another 3 weeks to do CHK-cache?  (They were the same 
 size beforehand).

Datastore resizing is slow. Especially if it's already mostly full. :(


signature.asc
Description: This is a digitally signed message part.
___
Support mailing list
Support@freenetproject.org
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe

Re: [freenet-support] Datastore resize

2012-09-28 Thread Dennis Nezic
On Fri, 28 Sep 2012 10:10:11 +0100, Matthew Toseland wrote:
 On Tuesday 25 Sep 2012 02:38:07 Pascal wrote:
  Changed my datastore size from 750g to 850g just over 5 hours
  ago. It now shows Datastore(CHK-store) resize in progress:
  108040/10717960. At this rate it will take about 3 weeks just to
  finish CHK-store. Does it them take another 3 weeks to do
  CHK-cache?  (They were the same size beforehand).
 
 Datastore resizing is slow. Especially if it's already mostly full. :(

3-weeks slow?
___
Support mailing list
Support@freenetproject.org
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe


Re: [freenet-support] Datastore resize

2012-09-28 Thread Pascal
With the 1st stage of the rebuild taking 3 weeks, and 1416 becoming 
mandatory in 2 weeks, if I had just waited my node would be unusable for 
a week.  I noticed an option to do the resize at startup instead that 
was supposed to be faster, so I set that and upgraded.  I let it run for 
a bit and then did the math, it was still going to take 3 weeks.  CPU 
and HDD were both practically idle.  I don't know what a resize does, 
but it sure seems like there is a bug in there somewhere.


-Pascal


On 9/28/2012 4:10 AM, Matthew Toseland wrote:

On Tuesday 25 Sep 2012 02:38:07 Pascal wrote:

Changed my datastore size from 750g to 850g just over 5 hours ago.
It now shows Datastore(CHK-store) resize in progress: 108040/10717960.
   At this rate it will take about 3 weeks just to finish CHK-store.
Does it them take another 3 weeks to do CHK-cache?  (They were the same
size beforehand).


Datastore resizing is slow. Especially if it's already mostly full. :(

___
Support mailing list
Support@freenetproject.org
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe


[freenet-support] Datastore resize

2012-09-24 Thread Pascal
Changed my datastore size from 750g to 850g just over 5 hours ago. 
It now shows Datastore(CHK-store) resize in progress: 108040/10717960. 
 At this rate it will take about 3 weeks just to finish CHK-store. 
Does it them take another 3 weeks to do CHK-cache?  (They were the same 
size beforehand).


-Pascal
___
Support mailing list
Support@freenetproject.org
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe


Re: [freenet-support] Datastore resize in loop-no problem any more, a question still

2010-10-16 Thread Matthew Toseland
On Tuesday 03 August 2010 14:06:12 Jep wrote:
 Jep schreef:
  My box seems to be big on FN loops!
  
  Increased the data store of Freenet from 40 to 50 gB and this
  'Datastore(CHK-cache) resize in progress: 159982/721390 ' keeps on 
  starting anew from zero, once the high number is reached.
  
  A restart of FN didn't help. What now? Restart winz?
  
  xp sp3, 2 gig memory, newest java.
 
 
 Hmm. This I caused myself, it's OK now. The resize process stops for 
 some time, and then starts again with other datastore parts, that made 
 me think it started over, so I tried a FN restart.
 
 Why does it take so long, now I'm writing anyway? It was chewing at it 
 all night.
 Making a new file using the cli command 'fsutil file createnew' takes 
 seconds only, no matter the size.

Because it has to move the data around, which takes a lot of seeking, because 
of how the datastore works.


signature.asc
Description: This is a digitally signed message part.
___
Support mailing list
Support@freenetproject.org
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe

Re: [freenet-support] Datastore resize in loop-no problem any more, a question still

2010-08-03 Thread Jep

Jep schreef:

My box seems to be big on FN loops!

Increased the data store of Freenet from 40 to 50 gB and this
'Datastore(CHK-cache) resize in progress: 159982/721390 ' keeps on 
starting anew from zero, once the high number is reached.


A restart of FN didn't help. What now? Restart winz?

xp sp3, 2 gig memory, newest java.



Hmm. This I caused myself, it's OK now. The resize process stops for 
some time, and then starts again with other datastore parts, that made 
me think it started over, so I tried a FN restart.


Why does it take so long, now I'm writing anyway? It was chewing at it 
all night.
Making a new file using the cli command 'fsutil file createnew' takes 
seconds only, no matter the size.


___
Support mailing list
Support@freenetproject.org
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe


[freenet-support] Datastore resize in loop

2010-08-02 Thread Jep

My box seems to be big on FN loops!

Increased the data store of Freenet from 40 to 50 gB and this
'Datastore(CHK-cache) resize in progress: 159982/721390 ' keeps on 
starting anew from zero, once the high number is reached.


A restart of FN didn't help. What now? Restart winz?

xp sp3, 2 gig memory, newest java.

___
Support mailing list
Support@freenetproject.org
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe