From: ceph-users [mailto:[email protected]] On Behalf Of 
[email protected]
Sent: 19 July 2016 07:44
To: Oliver Dzombic <[email protected]>; [email protected]
Subject: Re: [ceph-users] how to use cache tiering with proxy in ceph-10.2.2

 

I have configured ceph.conf with "osd_tier_promote_max_bytes_sec" in [osd] 
Attributes. But it still invalid.

I do command --show-config discovery that it has not been modified.

 

I don’t know why they are not getting picked up (did you restart the OSD’s), 
but you don’t want to do that anyway. Promoting too much will slow you down.

Note: Its caching tiering not just a cache. You only want stuff in cache if 
it’s hot.

 

 

[root@node01 ~]# cat /etc/ceph/ceph.conf | grep tier
osd_tier_promote_max_objects_sec=200000
osd_tier_promote_max_bytes_sec=16106127360

 

As above, way too high. Defaults are sensible, maybe 2x/4x if you need the 
cache to warm up quicker

 

 [root@node01 ~]# ceph --show-config | grep tier
mon_debug_unsafe_allow_tier_with_nonempty_snaps = false
osd_tier_promote_max_objects_sec = 5242880
osd_tier_promote_max_bytes_sec = 25
osd_tier_default_cache_mode = writeback
osd_tier_default_cache_hit_set_count = 4
osd_tier_default_cache_hit_set_period = 1200

 

Drop this to 60, unless your workload is very infrequent IO

 


osd_tier_default_cache_hit_set_type = bloom
osd_tier_default_cache_min_read_recency_for_promote = 1
osd_tier_default_cache_min_write_recency_for_promote = 1

 

Make these at least 2, otherwise you will promote on every IO


osd_tier_default_cache_hit_set_grade_decay_rate = 20
osd_tier_default_cache_hit_set_search_last_n = 1

 

and cache tiering does not work , low iops.

 

 

On Monday, July 18, 2016 5:33 PM, "[email protected] 
<mailto:[email protected]> " <[email protected] 
<mailto:[email protected]> > wrote:

 

thank you very much! 

 

On Monday, July 18, 2016 5:31 PM, Oliver Dzombic <[email protected] 
<mailto:[email protected]> > wrote:

 

Hi,

everything is here:

http://docs.ceph.com/docs/jewel/

except

osd_tier_promote_max_bytes_sec

and other stuff, but its enough there that you can make it work.

-- 
Mit freundlichen Gruessen / Best regards

Oliver Dzombic
IP-Interactive

mailto:[email protected] <mailto:[email protected]> 

Anschrift:

IP Interactive UG ( haftungsbeschraenkt )
Zum Sonnenberg 1-3
63571 Gelnhausen

HRB 93402 beim Amtsgericht Hanau
Geschäftsführung: Oliver Dzombic

Steuer Nr.: 35 236 3622 1
UST ID: DE274086107


Am 18.07.2016 um 11:24 schrieb [email protected] 
<mailto:[email protected]> :
> Where to find base docu?
> Official website does not update the document
> 
> 
> On Monday, July 18, 2016 5:16 PM, Oliver Dzombic
> <[email protected] <mailto:[email protected]> > wrote:
> 
> 
> Hi
> 
> i suggest you to read some base docu about that.
> 
> osd_tier_promote_max_bytes_sec = how much bytes per second are going on tier
> 
> ceph osd pool set ssd-pool target_max_bytes = maximum size in bytes on
> this specific pool ( its like a quota )
> 
> -- 
> Mit freundlichen Gruessen / Best regards
> 
> Oliver Dzombic
> IP-Interactive
> 
> mailto:[email protected] <mailto:[email protected]>  
> <mailto:[email protected] <mailto:[email protected]> >
> 
> Anschrift:
> 
> IP Interactive UG ( haftungsbeschraenkt )
> Zum Sonnenberg 1-3
> 63571 Gelnhausen
> 
> HRB 93402 beim Amtsgericht Hanau
> Geschäftsführung: Oliver Dzombic
> 
> Steuer Nr.: 35 236 3622 1
> UST ID: DE274086107
> 
> 
> Am 18.07.2016 um 11:14 schrieb [email protected] 
> <mailto:[email protected]> 
> <mailto:[email protected] <mailto:[email protected]> >:
>> what is "osd_tier_promote_max_bytes_sec" in ceph.conf file  and command
>> "ceph osd pool set ssd-pool target_max_bytes" are not the same ?
>>
>>
>> On Monday, July 18, 2016 4:40 PM, Oliver Dzombic
>> <[email protected] <mailto:[email protected]>  
>> <mailto:[email protected] <mailto:[email protected]> >> wrote:
>>
>>
>> Hi,
>>
>> osd_tier_promote_max_bytes_sec
>>
>> is your friend.
>>
>> --
>> Mit freundlichen Gruessen / Best regards
>>
>> Oliver Dzombic
>> IP-Interactive
>>
>> mailto:[email protected] <mailto:[email protected]>  
>> <mailto:[email protected] <mailto:[email protected]> >
> <mailto:[email protected] <mailto:[email protected]>  
> <mailto:[email protected] <mailto:[email protected]> >>
>>
>> Anschrift:
>>
>> IP Interactive UG ( haftungsbeschraenkt )
>> Zum Sonnenberg 1-3
>> 63571 Gelnhausen
>>
>> HRB 93402 beim Amtsgericht Hanau
>> Geschäftsführung: Oliver Dzombic
>>
>> Steuer Nr.: 35 236 3622 1
>> UST ID: DE274086107
>>
>>
>> Am 18.07.2016 um 10:19 schrieb [email protected] 
>> <mailto:[email protected]> 
> <mailto:[email protected] <mailto:[email protected]> >
>> <mailto:[email protected] <mailto:[email protected]>  
>> <mailto:[email protected] <mailto:[email protected]> >>:
>>> hello cepher!
>>>    I have a problem like this :
>>>    I want to config a cache tiering to my ceph with writeback mode.In
>>> ceph-0.94,it runs ok. IO is First through hot-pool. then it flush to
>>> cold-pool.
>>>    But in ceph-10.2.2,it doesn't like tihs. IO wrties to  hot-pool and
>>> cold-pool at the same time. I think it is be associated with "proxy".
>>>    So how to config a cache tiering in ceph-10.2.2 to Improve IO
>>> performance as much as possible ?
>>>    how to use the proxy in cache tiering ?
>>
>>>
>>>
>>> _______________________________________________
>>> ceph-users mailing list
>>> [email protected] <mailto:[email protected]>  
>>> <mailto:[email protected] <mailto:[email protected]> >
> <mailto:[email protected] <mailto:[email protected]>  
> <mailto:[email protected] <mailto:[email protected]> >>
>>> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
>>>
>> _______________________________________________
>> ceph-users mailing list
>> [email protected] <mailto:[email protected]>  
>> <mailto:[email protected] <mailto:[email protected]> >


> <mailto:[email protected] <mailto:[email protected]>  
> <mailto:[email protected] <mailto:[email protected]> >>
> 
>> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
>>
> _______________________________________________
> ceph-users mailing list
> [email protected] <mailto:[email protected]>  
> <mailto:[email protected] <mailto:[email protected]> >
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
> 
> 
_______________________________________________
ceph-users mailing list
[email protected] <mailto:[email protected]> 
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

 

 

_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

Reply via email to