Re: [ceph-users] about PG_Number

2015-11-17 Thread Vickie ch
By the way, here is a useful tool to calculate pg. http://ceph.com/pgcalc/ Best wishes, Mika 2015-11-18 11:46 GMT+08:00 Vickie ch : > Hi wah peng, > Hope you don't mind. Just for reference. > A extreme case. If your ceph cluster have 3 osd disks on different osd > server. > Set pg number is

Re: [ceph-users] about PG_Number

2015-11-17 Thread Vickie ch
Hi wah peng, Hope you don't mind. Just for reference. A extreme case. If your ceph cluster have 3 osd disks on different osd server. Set pg number is 10240.(Just example) That's mean all these pg will create on 3 disks. Lost one OSD also means a lot of pg lost too. It may bring some trouble for r

Re: [ceph-users] about PG_Number

2015-11-13 Thread Oleksandr Natalenko
"Learning Ceph" book gives us the following formula: PGs = OSDs × 100 / (replicas × pools) Saying, you have 10 OSDs and 5 pools with replica 2, you get: PGs = 10 × 100 / (2 × 5) = 100 PGs (per pool) It is also advised to round PGs count up to nearest power of 2. In this case, to 128. In typ

Re: [ceph-users] about PG_Number

2015-11-13 Thread Somnath Roy
f Of Francois Lafont Sent: Friday, November 13, 2015 4:34 AM To: ceph-users@lists.ceph.com Subject: Re: [ceph-users] about PG_Number Hi, On 13/11/2015 09:13, Vickie ch wrote: > If you have a large amount of OSDs but less pg number. You will find > your data write unevenly. > Some OSD h

Re: [ceph-users] about PG_Number

2015-11-13 Thread Robert LeBlanc
In our experience too few PGS leads to non-uniform disk load as well. We had lots of blocked IO when many of the disks were idle because there were and equal number always at 100% utilized. Lots of blocked IO when OSDS restarted and scanning the huge PGS (data). Scrubs were also painful for similar

Re: [ceph-users] about PG_Number

2015-11-13 Thread Francois Lafont
Hi, On 13/11/2015 09:13, Vickie ch wrote: > If you have a large amount of OSDs but less pg number. You will find your > data write unevenly. > Some OSD have no change to write data. > In the other side, pg number too large but OSD number too small that have a > chance to cause data lost. Data lo

Re: [ceph-users] about PG_Number

2015-11-13 Thread wah peng
why data lost happens? thanks. On 2015/11/13 星期五 16:13, Vickie ch wrote: In the other side, pg number too large but OSD number too small that have a chance to cause data lost. ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com

Re: [ceph-users] about PG_Number

2015-11-13 Thread Vickie ch
Hi wah peng, Just a thought. If you have a large amount of OSDs but less pg number. You will find your data write unevenly. Some OSD have no change to write data. In the other side, pg number too large but OSD number too small that have a chance to cause data lost. Best wishes, Mika 2015-11

[ceph-users] about PG_Number

2015-11-12 Thread wah peng
Hello, what's the disadvantage if setup PG_Number too large or too small against OSD number? Thanks. ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com