Re: Best plan for recovery if disk holding monitor data dies

2012-02-13 Thread Sage Weil
On Mon, 13 Feb 2012, Mandell Degerness wrote: > What would you recommend as the best recovery strategy when the disk > holding the monitor data for a given monitor fails? > > Should I recover from a periodic back up and re-start the monitor or > should I treat it as a new monitor being added to th

Best plan for recovery if disk holding monitor data dies

2012-02-13 Thread Mandell Degerness
What would you recommend as the best recovery strategy when the disk holding the monitor data for a given monitor fails? Should I recover from a periodic back up and re-start the monitor or should I treat it as a new monitor being added to the cluster? I suspect the former, but i'd like confirmat

Which SSD method is better for performance?

2012-02-13 Thread Paul Pettigrew
G'day all About to commence an R&D eval of the Ceph platform having been impressed with the momentum achieved over the past 12mths. I have one question re design before rolling out to metal I will be using 1x SSD drive per storage server node (assume it is /dev/sdb for this discussion)

Re: Stale NFS file handle

2012-02-13 Thread Tommi Virtanen
2012/2/13 Székelyi Szabolcs : > I'm using Ceph 0.41 with the FUSE client. After a while I get stale NFS file > errors when trying to read a file or list a directory. Logs and scrubbing > doesn't show any errors or suspicious entries. After remounting the filesystem > either by restarting the cluste

Re: Stale NFS file handle

2012-02-13 Thread Székelyi Szabolcs
On 2012. February 13. 15:54:39 Sage Weil wrote: > On Tue, 14 Feb 2012, Székelyi Szabolcs wrote: > > No, there's no NFS in the picture. The OSDs' backend storage is on a > > local filesystem. I think it's the FUSE client telling me this. > > Okay, that sounds like a bug then. The two interesting t

ceph_setattr mask

2012-02-13 Thread Noah Watkins
Howdy, It looks like ceph_fs.h contains the mask flags (e.g. CEPH_SETATTR_MODE) used in ceph_setattr, but I do not see these flags in any header installed from .deb files (grep /usr/include/*). Am I missing a location? Should these flags be part of the installed headers? Thanks, -Noah-- To uns

Re: Stale NFS file handle

2012-02-13 Thread Sage Weil
On Tue, 14 Feb 2012, Székelyi Szabolcs wrote: > On 2012. February 13. 15:34:13 Sage Weil wrote: > > On Tue, 14 Feb 2012, Székelyi Szabolcs wrote: > > > I'm using Ceph 0.41 with the FUSE client. After a while I get stale NFS > > > file errors when trying to read a file or list a directory. Logs and

Re: Stale NFS file handle

2012-02-13 Thread Székelyi Szabolcs
On 2012. February 13. 15:34:13 Sage Weil wrote: > On Tue, 14 Feb 2012, Székelyi Szabolcs wrote: > > I'm using Ceph 0.41 with the FUSE client. After a while I get stale NFS > > file errors when trying to read a file or list a directory. Logs and > > scrubbing doesn't show any errors or suspicious en

Re: Stale NFS file handle

2012-02-13 Thread Sage Weil
On Tue, 14 Feb 2012, Székelyi Szabolcs wrote: > I'm using Ceph 0.41 with the FUSE client. After a while I get stale NFS file > errors when trying to read a file or list a directory. Logs and scrubbing > doesn't show any errors or suspicious entries. After remounting the > filesystem > either by

Stale NFS file handle

2012-02-13 Thread Székelyi Szabolcs
Hi, I'm using Ceph 0.41 with the FUSE client. After a while I get stale NFS file errors when trying to read a file or list a directory. Logs and scrubbing doesn't show any errors or suspicious entries. After remounting the filesystem either by restarting the cluster thus forcing the clients to

Re: OSD id in configuration file

2012-02-13 Thread Sage Weil
On Mon, 13 Feb 2012, Dyweni - Ceph-Devel wrote: > Hi, > > That doesn't make sense... Would you explain further? > > This page (http://ceph.newdream.net/wiki/Cluster_configuration) says the id > can be a number or a name, which applies to (mon|mds|ods).$id. The monitors and mds's are identified b

Re: OSD id in configuration file

2012-02-13 Thread Dyweni - Ceph-Devel
Hi, That doesn't make sense... Would you explain further? This page (http://ceph.newdream.net/wiki/Cluster_configuration) says the id can be a number or a name, which applies to (mon|mds|ods).$id. Will I have problems using SHA1 sums to uniquely identify each OSD in my cluster? (i.e. OSD.$s

Re: Re: Problem with inconsistent PG

2012-02-13 Thread Sage Weil
On Sun, 12 Feb 2012, Jens Rehpoehler wrote: > > > Hi Liste, > > > > > > today i've got another problem. > > > > > > ceph -w shows up with an inconsistent PG over night: > > > > > > 2012-02-10 08:38:48.701775pg v441251: 1982 pgs: 1981 active+clean, 1 > > > active+clean+inconsistent; 179

Re: Problem with inconsistent PG

2012-02-13 Thread Sage Weil
On Sun, 12 Feb 2012, Jens Rehpoehler wrote: > Am 12.02.2012 13:00, schrieb Jens Rehpoehler: > > > > Hi Liste, > > > > > > > > today i've got another problem. > > > > > > > > ceph -w shows up with an inconsistent PG over night: > > > > > > > > 2012-02-10 08:38:48.701775pg v441251: 1982 pg

Re: OSD id in configuration file

2012-02-13 Thread Sage Weil
On Mon, 13 Feb 2012, eric_yh_c...@wistron.com wrote: > Hi, all: > > For the scalability consideration, we would like to name the first > harddisk as "00101" on first server. > > And named the first harddisk as "00201" on second server. The ceph.conf > seems like this: > > [osd] > osd data = /srv

Re: [RFC PATCH 0/6] Understanding delays due to throttling under very heavy write load

2012-02-13 Thread Jim Schutt
On 02/10/2012 05:05 PM, sridhar basam wrote: > But the server never ACKed that packet. Too busy? > > I was collecting vmstat data during the run; here's the important bits: > > Fri Feb 10 11:56:51 MST 2012 > vmstat -w 8 16 > procs ---memory-- ---swap-- -i

OSD id in configuration file

2012-02-13 Thread Eric_YH_Chen
Hi, all: For the scalability consideration, we would like to name the first harddisk as "00101" on first server. And named the first harddisk as "00201" on second server. The ceph.conf seems like this: [osd] osd data = /srv/osd.$id osd journal = /srv/osd.$id.journal osd journal size = 1000 [osd