| Message ID | 20211020143708.14728-1-lhenriques@suse.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [RFC] ceph: add remote object copy counter to fs client | expand |
On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > This counter will keep track of the number of remote object copies done on > copy_file_range syscalls. This counter will be filesystem per-client, and > can be accessed from the client debugfs directory. > > Cc: Patrick Donnelly <pdonnell@redhat.com> > Signed-off-by: Luís Henriques <lhenriques@suse.de> > --- > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > 100% sure about the usefulness of this patch, or if this is the best way > to provide the functionality Patrick requested. Anyway, this is just to > get some feedback, hence the RFC. > > Cheers, > -- > Luís > > [0] https://github.com/ceph/ceph/pull/42720 > I think this would be better integrated into the stats infrastructure. Maybe you could add a new set of "copy" stats to struct ceph_client_metric that tracks the total copy operations done, their size and latency (similar to read and write ops)? > fs/ceph/debugfs.c | 17 ++++++++++++++++- > fs/ceph/file.c | 1 + > fs/ceph/super.c | 1 + > fs/ceph/super.h | 2 ++ > 4 files changed, 20 insertions(+), 1 deletion(-) > > diff --git a/fs/ceph/debugfs.c b/fs/ceph/debugfs.c > index 38b78b45811f..09f4c04ade0e 100644 > --- a/fs/ceph/debugfs.c > +++ b/fs/ceph/debugfs.c > @@ -346,13 +346,22 @@ static int status_show(struct seq_file *s, void *p) > return 0; > } > > +static int copyfrom_show(struct seq_file *s, void *p) > +{ > + struct ceph_fs_client *fsc = s->private; > + > + seq_printf(s, "%llu\n", atomic64_read(&fsc->copyfrom_count)); > + > + return 0; > +} > + > DEFINE_SHOW_ATTRIBUTE(mdsmap); > DEFINE_SHOW_ATTRIBUTE(mdsc); > DEFINE_SHOW_ATTRIBUTE(caps); > DEFINE_SHOW_ATTRIBUTE(mds_sessions); > DEFINE_SHOW_ATTRIBUTE(metric); > DEFINE_SHOW_ATTRIBUTE(status); > - > +DEFINE_SHOW_ATTRIBUTE(copyfrom); > > /* > * debugfs > @@ -387,6 +396,7 @@ void ceph_fs_debugfs_cleanup(struct ceph_fs_client *fsc) > debugfs_remove(fsc->debugfs_caps); > debugfs_remove(fsc->debugfs_metric); > debugfs_remove(fsc->debugfs_mdsc); > + debugfs_remove(fsc->debugfs_copyfrom); > } > > void ceph_fs_debugfs_init(struct ceph_fs_client *fsc) > @@ -443,6 +453,11 @@ void ceph_fs_debugfs_init(struct ceph_fs_client *fsc) > fsc->client->debugfs_dir, > fsc, > &status_fops); > + fsc->debugfs_copyfrom = debugfs_create_file("copyfrom", > + 0400, > + fsc->client->debugfs_dir, > + fsc, > + ©from_fops); > } > > > diff --git a/fs/ceph/file.c b/fs/ceph/file.c > index d16fd2d5fd42..bbeb437ca4bf 100644 > --- a/fs/ceph/file.c > +++ b/fs/ceph/file.c > @@ -2254,6 +2254,7 @@ static ssize_t ceph_do_objects_copy(struct ceph_inode_info *src_ci, u64 *src_off > bytes = ret; > goto out; > } > + atomic64_inc(&fsc->copyfrom_count); > len -= object_size; > bytes += object_size; > *src_off += object_size; > diff --git a/fs/ceph/super.c b/fs/ceph/super.c > index 9b1b7f4cfdd4..4972554185e3 100644 > --- a/fs/ceph/super.c > +++ b/fs/ceph/super.c > @@ -670,6 +670,7 @@ static struct ceph_fs_client *create_fs_client(struct ceph_mount_options *fsopt, > fsc->have_copy_from2 = true; > > atomic_long_set(&fsc->writeback_count, 0); > + atomic64_set(&fsc->copyfrom_count, 0); > > err = -ENOMEM; > /* > diff --git a/fs/ceph/super.h b/fs/ceph/super.h > index a40eb14c282a..65846beca418 100644 > --- a/fs/ceph/super.h > +++ b/fs/ceph/super.h > @@ -119,6 +119,7 @@ struct ceph_fs_client { > struct ceph_mds_client *mdsc; > > atomic_long_t writeback_count; > + atomic64_t copyfrom_count; > > struct workqueue_struct *inode_wq; > struct workqueue_struct *cap_wq; > @@ -131,6 +132,7 @@ struct ceph_fs_client { > struct dentry *debugfs_metric; > struct dentry *debugfs_status; > struct dentry *debugfs_mds_sessions; > + struct dentry *debugfs_copyfrom; > #endif > > #ifdef CONFIG_CEPH_FSCACHE
On Wed, Oct 20, 2021 at 12:27:21PM -0400, Jeff Layton wrote: > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > This counter will keep track of the number of remote object copies done on > > copy_file_range syscalls. This counter will be filesystem per-client, and > > can be accessed from the client debugfs directory. > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > --- > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > 100% sure about the usefulness of this patch, or if this is the best way > > to provide the functionality Patrick requested. Anyway, this is just to > > get some feedback, hence the RFC. > > > > Cheers, > > -- > > Luís > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > I think this would be better integrated into the stats infrastructure. > > Maybe you could add a new set of "copy" stats to struct > ceph_client_metric that tracks the total copy operations done, their > size and latency (similar to read and write ops)? Ah, yeah. That probably makes sense, I'll have a closer look at that infrastructure. I suspect that involves touching the MDS code, which means that unfortunately it'll take me a bit longer to send out v2. Cheers, -- Luís > > fs/ceph/debugfs.c | 17 ++++++++++++++++- > > fs/ceph/file.c | 1 + > > fs/ceph/super.c | 1 + > > fs/ceph/super.h | 2 ++ > > 4 files changed, 20 insertions(+), 1 deletion(-) > > > > diff --git a/fs/ceph/debugfs.c b/fs/ceph/debugfs.c > > index 38b78b45811f..09f4c04ade0e 100644 > > --- a/fs/ceph/debugfs.c > > +++ b/fs/ceph/debugfs.c > > @@ -346,13 +346,22 @@ static int status_show(struct seq_file *s, void *p) > > return 0; > > } > > > > +static int copyfrom_show(struct seq_file *s, void *p) > > +{ > > + struct ceph_fs_client *fsc = s->private; > > + > > + seq_printf(s, "%llu\n", atomic64_read(&fsc->copyfrom_count)); > > + > > + return 0; > > +} > > + > > DEFINE_SHOW_ATTRIBUTE(mdsmap); > > DEFINE_SHOW_ATTRIBUTE(mdsc); > > DEFINE_SHOW_ATTRIBUTE(caps); > > DEFINE_SHOW_ATTRIBUTE(mds_sessions); > > DEFINE_SHOW_ATTRIBUTE(metric); > > DEFINE_SHOW_ATTRIBUTE(status); > > - > > +DEFINE_SHOW_ATTRIBUTE(copyfrom); > > > > /* > > * debugfs > > @@ -387,6 +396,7 @@ void ceph_fs_debugfs_cleanup(struct ceph_fs_client *fsc) > > debugfs_remove(fsc->debugfs_caps); > > debugfs_remove(fsc->debugfs_metric); > > debugfs_remove(fsc->debugfs_mdsc); > > + debugfs_remove(fsc->debugfs_copyfrom); > > } > > > > void ceph_fs_debugfs_init(struct ceph_fs_client *fsc) > > @@ -443,6 +453,11 @@ void ceph_fs_debugfs_init(struct ceph_fs_client *fsc) > > fsc->client->debugfs_dir, > > fsc, > > &status_fops); > > + fsc->debugfs_copyfrom = debugfs_create_file("copyfrom", > > + 0400, > > + fsc->client->debugfs_dir, > > + fsc, > > + ©from_fops); > > } > > > > > > diff --git a/fs/ceph/file.c b/fs/ceph/file.c > > index d16fd2d5fd42..bbeb437ca4bf 100644 > > --- a/fs/ceph/file.c > > +++ b/fs/ceph/file.c > > @@ -2254,6 +2254,7 @@ static ssize_t ceph_do_objects_copy(struct ceph_inode_info *src_ci, u64 *src_off > > bytes = ret; > > goto out; > > } > > + atomic64_inc(&fsc->copyfrom_count); > > len -= object_size; > > bytes += object_size; > > *src_off += object_size; > > diff --git a/fs/ceph/super.c b/fs/ceph/super.c > > index 9b1b7f4cfdd4..4972554185e3 100644 > > --- a/fs/ceph/super.c > > +++ b/fs/ceph/super.c > > @@ -670,6 +670,7 @@ static struct ceph_fs_client *create_fs_client(struct ceph_mount_options *fsopt, > > fsc->have_copy_from2 = true; > > > > atomic_long_set(&fsc->writeback_count, 0); > > + atomic64_set(&fsc->copyfrom_count, 0); > > > > err = -ENOMEM; > > /* > > diff --git a/fs/ceph/super.h b/fs/ceph/super.h > > index a40eb14c282a..65846beca418 100644 > > --- a/fs/ceph/super.h > > +++ b/fs/ceph/super.h > > @@ -119,6 +119,7 @@ struct ceph_fs_client { > > struct ceph_mds_client *mdsc; > > > > atomic_long_t writeback_count; > > + atomic64_t copyfrom_count; > > > > struct workqueue_struct *inode_wq; > > struct workqueue_struct *cap_wq; > > @@ -131,6 +132,7 @@ struct ceph_fs_client { > > struct dentry *debugfs_metric; > > struct dentry *debugfs_status; > > struct dentry *debugfs_mds_sessions; > > + struct dentry *debugfs_copyfrom; > > #endif > > > > #ifdef CONFIG_CEPH_FSCACHE > > -- > Jeff Layton <jlayton@kernel.org> >
On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: > > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > This counter will keep track of the number of remote object copies done on > > copy_file_range syscalls. This counter will be filesystem per-client, and > > can be accessed from the client debugfs directory. > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > --- > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > 100% sure about the usefulness of this patch, or if this is the best way > > to provide the functionality Patrick requested. Anyway, this is just to > > get some feedback, hence the RFC. > > > > Cheers, > > -- > > Luís > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > I think this would be better integrated into the stats infrastructure. > > Maybe you could add a new set of "copy" stats to struct > ceph_client_metric that tracks the total copy operations done, their > size and latency (similar to read and write ops)? I think it's a good idea to integrate this into "stats" but I think a local debugfs file for some counters is still useful. The "stats" module is immature at this time and I'd rather not build any qa tests (yet) that rely on it. Can we generalize this patch-set to a file named "op_counters" or similar and additionally add other OSD ops performed by the kclient?
On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: > On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > > This counter will keep track of the number of remote object copies done on > > > copy_file_range syscalls. This counter will be filesystem per-client, and > > > can be accessed from the client debugfs directory. > > > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > > --- > > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > > 100% sure about the usefulness of this patch, or if this is the best way > > > to provide the functionality Patrick requested. Anyway, this is just to > > > get some feedback, hence the RFC. > > > > > > Cheers, > > > -- > > > Luís > > > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > > > > I think this would be better integrated into the stats infrastructure. > > > > Maybe you could add a new set of "copy" stats to struct > > ceph_client_metric that tracks the total copy operations done, their > > size and latency (similar to read and write ops)? > > I think it's a good idea to integrate this into "stats" but I think a > local debugfs file for some counters is still useful. The "stats" > module is immature at this time and I'd rather not build any qa tests > (yet) that rely on it. > > Can we generalize this patch-set to a file named "op_counters" or > similar and additionally add other OSD ops performed by the kclient? > Tracking this sort of thing is the main purpose of the stats code. I'm really not keen on adding a whole separate set of files for reporting this. What's the specific problem with relying on the data in debugfs "metrics" file?
On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: > > On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: > > On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > > > This counter will keep track of the number of remote object copies done on > > > > copy_file_range syscalls. This counter will be filesystem per-client, and > > > > can be accessed from the client debugfs directory. > > > > > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > > > --- > > > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > > > 100% sure about the usefulness of this patch, or if this is the best way > > > > to provide the functionality Patrick requested. Anyway, this is just to > > > > get some feedback, hence the RFC. > > > > > > > > Cheers, > > > > -- > > > > Luís > > > > > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > > > > > > > I think this would be better integrated into the stats infrastructure. > > > > > > Maybe you could add a new set of "copy" stats to struct > > > ceph_client_metric that tracks the total copy operations done, their > > > size and latency (similar to read and write ops)? > > > > I think it's a good idea to integrate this into "stats" but I think a > > local debugfs file for some counters is still useful. The "stats" > > module is immature at this time and I'd rather not build any qa tests > > (yet) that rely on it. > > > > Can we generalize this patch-set to a file named "op_counters" or > > similar and additionally add other OSD ops performed by the kclient? > > > > > Tracking this sort of thing is the main purpose of the stats code. I'm > really not keen on adding a whole separate set of files for reporting > this. Maybe I'm confused. Is there some "file" which is already used for this type of debugging information? Or do you mean the code for sending stats to the MDS to support cephfs-top? > What's the specific problem with relying on the data in debugfs > "metrics" file? Maybe no problem? I wasn't aware of a "metrics" file.
On Thu, 2021-10-21 at 12:18 -0400, Patrick Donnelly wrote: > On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: > > > > On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: > > > On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > > > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > > > > This counter will keep track of the number of remote object copies done on > > > > > copy_file_range syscalls. This counter will be filesystem per-client, and > > > > > can be accessed from the client debugfs directory. > > > > > > > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > > > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > > > > --- > > > > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > > > > 100% sure about the usefulness of this patch, or if this is the best way > > > > > to provide the functionality Patrick requested. Anyway, this is just to > > > > > get some feedback, hence the RFC. > > > > > > > > > > Cheers, > > > > > -- > > > > > Luís > > > > > > > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > > > > > > > > > > I think this would be better integrated into the stats infrastructure. > > > > > > > > Maybe you could add a new set of "copy" stats to struct > > > > ceph_client_metric that tracks the total copy operations done, their > > > > size and latency (similar to read and write ops)? > > > > > > I think it's a good idea to integrate this into "stats" but I think a > > > local debugfs file for some counters is still useful. The "stats" > > > module is immature at this time and I'd rather not build any qa tests > > > (yet) that rely on it. > > > > > > Can we generalize this patch-set to a file named "op_counters" or > > > similar and additionally add other OSD ops performed by the kclient? > > > > > > > > > Tracking this sort of thing is the main purpose of the stats code. I'm > > really not keen on adding a whole separate set of files for reporting > > this. > > Maybe I'm confused. Is there some "file" which is already used for > this type of debugging information? Or do you mean the code for > sending stats to the MDS to support cephfs-top? > > > What's the specific problem with relying on the data in debugfs > > "metrics" file? > > Maybe no problem? I wasn't aware of a "metrics" file. > Yes. For instance: # cat /sys/kernel/debug/ceph/*/metrics item total ------------------------------------------ opened files / total inodes 0 / 4 pinned i_caps / total inodes 5 / 4 opened inodes / total inodes 0 / 4 item total avg_lat(us) min_lat(us) max_lat(us) stdev(us) ----------------------------------------------------------------------------------- read 0 0 0 0 0 write 5 914013 824797 1092343 103476 metadata 79 12856 1572 114572 13262 item total avg_sz(bytes) min_sz(bytes) max_sz(bytes) total_sz(bytes) ---------------------------------------------------------------------------------------- read 0 0 0 0 0 write 5 4194304 4194304 4194304 20971520 item total miss hit ------------------------------------------------- d_lease 11 0 29 caps 5 68 10702 I'm proposing that Luis add new lines for "copy" to go along with the "read" and "write" ones. The "total" counter should give you a count of the number of operations.
On Thu, Oct 21, 2021 at 12:35 PM Jeff Layton <jlayton@kernel.org> wrote: > > On Thu, 2021-10-21 at 12:18 -0400, Patrick Donnelly wrote: > > On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: > > > > On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > > > > > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > > > > > This counter will keep track of the number of remote object copies done on > > > > > > copy_file_range syscalls. This counter will be filesystem per-client, and > > > > > > can be accessed from the client debugfs directory. > > > > > > > > > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > > > > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > > > > > --- > > > > > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > > > > > 100% sure about the usefulness of this patch, or if this is the best way > > > > > > to provide the functionality Patrick requested. Anyway, this is just to > > > > > > get some feedback, hence the RFC. > > > > > > > > > > > > Cheers, > > > > > > -- > > > > > > Luís > > > > > > > > > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > > > > > > > > > > > > > I think this would be better integrated into the stats infrastructure. > > > > > > > > > > Maybe you could add a new set of "copy" stats to struct > > > > > ceph_client_metric that tracks the total copy operations done, their > > > > > size and latency (similar to read and write ops)? > > > > > > > > I think it's a good idea to integrate this into "stats" but I think a > > > > local debugfs file for some counters is still useful. The "stats" > > > > module is immature at this time and I'd rather not build any qa tests > > > > (yet) that rely on it. > > > > > > > > Can we generalize this patch-set to a file named "op_counters" or > > > > similar and additionally add other OSD ops performed by the kclient? > > > > > > > > > > > > > Tracking this sort of thing is the main purpose of the stats code. I'm > > > really not keen on adding a whole separate set of files for reporting > > > this. > > > > Maybe I'm confused. Is there some "file" which is already used for > > this type of debugging information? Or do you mean the code for > > sending stats to the MDS to support cephfs-top? > > > > > What's the specific problem with relying on the data in debugfs > > > "metrics" file? > > > > Maybe no problem? I wasn't aware of a "metrics" file. > > > > Yes. For instance: > > # cat /sys/kernel/debug/ceph/*/metrics > item total > ------------------------------------------ > opened files / total inodes 0 / 4 > pinned i_caps / total inodes 5 / 4 > opened inodes / total inodes 0 / 4 > > item total avg_lat(us) min_lat(us) max_lat(us) stdev(us) > ----------------------------------------------------------------------------------- > read 0 0 0 0 0 > write 5 914013 824797 1092343 103476 > metadata 79 12856 1572 114572 13262 > > item total avg_sz(bytes) min_sz(bytes) max_sz(bytes) total_sz(bytes) > ---------------------------------------------------------------------------------------- > read 0 0 0 0 0 > write 5 4194304 4194304 4194304 20971520 > > item total miss hit > ------------------------------------------------- > d_lease 11 0 29 > caps 5 68 10702 > > > I'm proposing that Luis add new lines for "copy" to go along with the > "read" and "write" ones. The "total" counter should give you a count of > the number of operations. Okay that makes more sense! Side note: I am a bit horrified by how computer-unfriendly that table-formatted data is.
On Thu, 2021-10-21 at 13:30 -0400, Patrick Donnelly wrote: > On Thu, Oct 21, 2021 at 12:35 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > On Thu, 2021-10-21 at 12:18 -0400, Patrick Donnelly wrote: > > > On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > > > On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: > > > > > On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > > > > > > > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > > > > > > This counter will keep track of the number of remote object copies done on > > > > > > > copy_file_range syscalls. This counter will be filesystem per-client, and > > > > > > > can be accessed from the client debugfs directory. > > > > > > > > > > > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > > > > > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > > > > > > --- > > > > > > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > > > > > > 100% sure about the usefulness of this patch, or if this is the best way > > > > > > > to provide the functionality Patrick requested. Anyway, this is just to > > > > > > > get some feedback, hence the RFC. > > > > > > > > > > > > > > Cheers, > > > > > > > -- > > > > > > > Luís > > > > > > > > > > > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > > > > > > > > > > > > > > > > I think this would be better integrated into the stats infrastructure. > > > > > > > > > > > > Maybe you could add a new set of "copy" stats to struct > > > > > > ceph_client_metric that tracks the total copy operations done, their > > > > > > size and latency (similar to read and write ops)? > > > > > > > > > > I think it's a good idea to integrate this into "stats" but I think a > > > > > local debugfs file for some counters is still useful. The "stats" > > > > > module is immature at this time and I'd rather not build any qa tests > > > > > (yet) that rely on it. > > > > > > > > > > Can we generalize this patch-set to a file named "op_counters" or > > > > > similar and additionally add other OSD ops performed by the kclient? > > > > > > > > > > > > > > > > > Tracking this sort of thing is the main purpose of the stats code. I'm > > > > really not keen on adding a whole separate set of files for reporting > > > > this. > > > > > > Maybe I'm confused. Is there some "file" which is already used for > > > this type of debugging information? Or do you mean the code for > > > sending stats to the MDS to support cephfs-top? > > > > > > > What's the specific problem with relying on the data in debugfs > > > > "metrics" file? > > > > > > Maybe no problem? I wasn't aware of a "metrics" file. > > > > > > > Yes. For instance: > > > > # cat /sys/kernel/debug/ceph/*/metrics > > item total > > ------------------------------------------ > > opened files / total inodes 0 / 4 > > pinned i_caps / total inodes 5 / 4 > > opened inodes / total inodes 0 / 4 > > > > item total avg_lat(us) min_lat(us) max_lat(us) stdev(us) > > ----------------------------------------------------------------------------------- > > read 0 0 0 0 0 > > write 5 914013 824797 1092343 103476 > > metadata 79 12856 1572 114572 13262 > > > > item total avg_sz(bytes) min_sz(bytes) max_sz(bytes) total_sz(bytes) > > ---------------------------------------------------------------------------------------- > > read 0 0 0 0 0 > > write 5 4194304 4194304 4194304 20971520 > > > > item total miss hit > > ------------------------------------------------- > > d_lease 11 0 29 > > caps 5 68 10702 > > > > > > I'm proposing that Luis add new lines for "copy" to go along with the > > "read" and "write" ones. The "total" counter should give you a count of > > the number of operations. > > Okay that makes more sense! > > Side note: I am a bit horrified by how computer-unfriendly that > table-formatted data is. > Not going to disagree with you there. I'm happy to consider patches to reformat that to be more machine-readable.
On Thu, Oct 21, 2021 at 12:35:18PM -0400, Jeff Layton wrote: > On Thu, 2021-10-21 at 12:18 -0400, Patrick Donnelly wrote: > > On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: > > > > On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > > > > > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > > > > > This counter will keep track of the number of remote object copies done on > > > > > > copy_file_range syscalls. This counter will be filesystem per-client, and > > > > > > can be accessed from the client debugfs directory. > > > > > > > > > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > > > > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > > > > > --- > > > > > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > > > > > 100% sure about the usefulness of this patch, or if this is the best way > > > > > > to provide the functionality Patrick requested. Anyway, this is just to > > > > > > get some feedback, hence the RFC. > > > > > > > > > > > > Cheers, > > > > > > -- > > > > > > Luís > > > > > > > > > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > > > > > > > > > > > > > I think this would be better integrated into the stats infrastructure. > > > > > > > > > > Maybe you could add a new set of "copy" stats to struct > > > > > ceph_client_metric that tracks the total copy operations done, their > > > > > size and latency (similar to read and write ops)? > > > > > > > > I think it's a good idea to integrate this into "stats" but I think a > > > > local debugfs file for some counters is still useful. The "stats" > > > > module is immature at this time and I'd rather not build any qa tests > > > > (yet) that rely on it. > > > > > > > > Can we generalize this patch-set to a file named "op_counters" or > > > > similar and additionally add other OSD ops performed by the kclient? > > > > > > > > > > > > > Tracking this sort of thing is the main purpose of the stats code. I'm > > > really not keen on adding a whole separate set of files for reporting > > > this. > > > > Maybe I'm confused. Is there some "file" which is already used for > > this type of debugging information? Or do you mean the code for > > sending stats to the MDS to support cephfs-top? > > > > > What's the specific problem with relying on the data in debugfs > > > "metrics" file? > > > > Maybe no problem? I wasn't aware of a "metrics" file. > > > > Yes. For instance: > > # cat /sys/kernel/debug/ceph/*/metrics > item total > ------------------------------------------ > opened files / total inodes 0 / 4 > pinned i_caps / total inodes 5 / 4 > opened inodes / total inodes 0 / 4 > > item total avg_lat(us) min_lat(us) max_lat(us) stdev(us) > ----------------------------------------------------------------------------------- > read 0 0 0 0 0 > write 5 914013 824797 1092343 103476 > metadata 79 12856 1572 114572 13262 > > item total avg_sz(bytes) min_sz(bytes) max_sz(bytes) total_sz(bytes) > ---------------------------------------------------------------------------------------- > read 0 0 0 0 0 > write 5 4194304 4194304 4194304 20971520 > > item total miss hit > ------------------------------------------------- > d_lease 11 0 29 > caps 5 68 10702 > > > I'm proposing that Luis add new lines for "copy" to go along with the > "read" and "write" ones. The "total" counter should give you a count of > the number of operations. The problem with this is that it will require quite some work on the MDS-side because, AFAIU, the MDS will need to handle different versions of the CEPH_MSG_CLIENT_METRICS message (with and without the new copy-from metrics). Will this extra metric ever be useful on the MDS side? From what I understood Patrick's initial request was to have a way to find out, on the client, if remote copies are really happening. (*sigh* for not having tracepoints.) Anyway, I can look into adding this to the metrics infrastructure, but it'll likely take me some more time to get to it and to figure out (once again) how the messages versioning work. Cheers, -- Luís
On Mon, 2021-10-25 at 11:12 +0100, Luís Henriques wrote: > On Thu, Oct 21, 2021 at 12:35:18PM -0400, Jeff Layton wrote: > > On Thu, 2021-10-21 at 12:18 -0400, Patrick Donnelly wrote: > > > On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > > > On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: > > > > > On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > > > > > > > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > > > > > > This counter will keep track of the number of remote object copies done on > > > > > > > copy_file_range syscalls. This counter will be filesystem per-client, and > > > > > > > can be accessed from the client debugfs directory. > > > > > > > > > > > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > > > > > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > > > > > > --- > > > > > > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > > > > > > 100% sure about the usefulness of this patch, or if this is the best way > > > > > > > to provide the functionality Patrick requested. Anyway, this is just to > > > > > > > get some feedback, hence the RFC. > > > > > > > > > > > > > > Cheers, > > > > > > > -- > > > > > > > Luís > > > > > > > > > > > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > > > > > > > > > > > > > > > > I think this would be better integrated into the stats infrastructure. > > > > > > > > > > > > Maybe you could add a new set of "copy" stats to struct > > > > > > ceph_client_metric that tracks the total copy operations done, their > > > > > > size and latency (similar to read and write ops)? > > > > > > > > > > I think it's a good idea to integrate this into "stats" but I think a > > > > > local debugfs file for some counters is still useful. The "stats" > > > > > module is immature at this time and I'd rather not build any qa tests > > > > > (yet) that rely on it. > > > > > > > > > > Can we generalize this patch-set to a file named "op_counters" or > > > > > similar and additionally add other OSD ops performed by the kclient? > > > > > > > > > > > > > > > > > Tracking this sort of thing is the main purpose of the stats code. I'm > > > > really not keen on adding a whole separate set of files for reporting > > > > this. > > > > > > Maybe I'm confused. Is there some "file" which is already used for > > > this type of debugging information? Or do you mean the code for > > > sending stats to the MDS to support cephfs-top? > > > > > > > What's the specific problem with relying on the data in debugfs > > > > "metrics" file? > > > > > > Maybe no problem? I wasn't aware of a "metrics" file. > > > > > > > Yes. For instance: > > > > # cat /sys/kernel/debug/ceph/*/metrics > > item total > > ------------------------------------------ > > opened files / total inodes 0 / 4 > > pinned i_caps / total inodes 5 / 4 > > opened inodes / total inodes 0 / 4 > > > > item total avg_lat(us) min_lat(us) max_lat(us) stdev(us) > > ----------------------------------------------------------------------------------- > > read 0 0 0 0 0 > > write 5 914013 824797 1092343 103476 > > metadata 79 12856 1572 114572 13262 > > > > item total avg_sz(bytes) min_sz(bytes) max_sz(bytes) total_sz(bytes) > > ---------------------------------------------------------------------------------------- > > read 0 0 0 0 0 > > write 5 4194304 4194304 4194304 20971520 > > > > item total miss hit > > ------------------------------------------------- > > d_lease 11 0 29 > > caps 5 68 10702 > > > > > > I'm proposing that Luis add new lines for "copy" to go along with the > > "read" and "write" ones. The "total" counter should give you a count of > > the number of operations. > > The problem with this is that it will require quite some work on the > MDS-side because, AFAIU, the MDS will need to handle different versions of > the CEPH_MSG_CLIENT_METRICS message (with and without the new copy-from > metrics). > > Will this extra metric ever be useful on the MDS side? From what I > understood Patrick's initial request was to have a way to find out, on the > client, if remote copies are really happening. (*sigh* for not having > tracepoints.) > > Anyway, I can look into adding this to the metrics infrastructure, but > it'll likely take me some more time to get to it and to figure out (once > again) how the messages versioning work. > I think it is useful info to report to the MDS, but it's not required to send these to the MDS to solve the current problem. My suggestion would be to add what's needed to track these stats in the kclient and report them via debugfs, but don't send the info to the MDS just yet. Later, we could extend the protocol with COPY stats, and add the necessary infrastructure to the MDS to deal with it. Once that's in place, we can then extend the kclient to start sending this info along when it reports the stats.
On Mon, Oct 25, 2021 at 06:20:40AM -0400, Jeff Layton wrote: > On Mon, 2021-10-25 at 11:12 +0100, Luís Henriques wrote: > > On Thu, Oct 21, 2021 at 12:35:18PM -0400, Jeff Layton wrote: > > > On Thu, 2021-10-21 at 12:18 -0400, Patrick Donnelly wrote: > > > > On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > > > > > On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: > > > > > > On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > > > > > > > > > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > > > > > > > This counter will keep track of the number of remote object copies done on > > > > > > > > copy_file_range syscalls. This counter will be filesystem per-client, and > > > > > > > > can be accessed from the client debugfs directory. > > > > > > > > > > > > > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > > > > > > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > > > > > > > --- > > > > > > > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > > > > > > > 100% sure about the usefulness of this patch, or if this is the best way > > > > > > > > to provide the functionality Patrick requested. Anyway, this is just to > > > > > > > > get some feedback, hence the RFC. > > > > > > > > > > > > > > > > Cheers, > > > > > > > > -- > > > > > > > > Luís > > > > > > > > > > > > > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > > > > > > > > > > > > > > > > > > > I think this would be better integrated into the stats infrastructure. > > > > > > > > > > > > > > Maybe you could add a new set of "copy" stats to struct > > > > > > > ceph_client_metric that tracks the total copy operations done, their > > > > > > > size and latency (similar to read and write ops)? > > > > > > > > > > > > I think it's a good idea to integrate this into "stats" but I think a > > > > > > local debugfs file for some counters is still useful. The "stats" > > > > > > module is immature at this time and I'd rather not build any qa tests > > > > > > (yet) that rely on it. > > > > > > > > > > > > Can we generalize this patch-set to a file named "op_counters" or > > > > > > similar and additionally add other OSD ops performed by the kclient? > > > > > > > > > > > > > > > > > > > > > Tracking this sort of thing is the main purpose of the stats code. I'm > > > > > really not keen on adding a whole separate set of files for reporting > > > > > this. > > > > > > > > Maybe I'm confused. Is there some "file" which is already used for > > > > this type of debugging information? Or do you mean the code for > > > > sending stats to the MDS to support cephfs-top? > > > > > > > > > What's the specific problem with relying on the data in debugfs > > > > > "metrics" file? > > > > > > > > Maybe no problem? I wasn't aware of a "metrics" file. > > > > > > > > > > Yes. For instance: > > > > > > # cat /sys/kernel/debug/ceph/*/metrics > > > item total > > > ------------------------------------------ > > > opened files / total inodes 0 / 4 > > > pinned i_caps / total inodes 5 / 4 > > > opened inodes / total inodes 0 / 4 > > > > > > item total avg_lat(us) min_lat(us) max_lat(us) stdev(us) > > > ----------------------------------------------------------------------------------- > > > read 0 0 0 0 0 > > > write 5 914013 824797 1092343 103476 > > > metadata 79 12856 1572 114572 13262 > > > > > > item total avg_sz(bytes) min_sz(bytes) max_sz(bytes) total_sz(bytes) > > > ---------------------------------------------------------------------------------------- > > > read 0 0 0 0 0 > > > write 5 4194304 4194304 4194304 20971520 > > > > > > item total miss hit > > > ------------------------------------------------- > > > d_lease 11 0 29 > > > caps 5 68 10702 > > > > > > > > > I'm proposing that Luis add new lines for "copy" to go along with the > > > "read" and "write" ones. The "total" counter should give you a count of > > > the number of operations. > > > > The problem with this is that it will require quite some work on the > > MDS-side because, AFAIU, the MDS will need to handle different versions of > > the CEPH_MSG_CLIENT_METRICS message (with and without the new copy-from > > metrics). > > > > Will this extra metric ever be useful on the MDS side? From what I > > understood Patrick's initial request was to have a way to find out, on the > > client, if remote copies are really happening. (*sigh* for not having > > tracepoints.) > > > > Anyway, I can look into adding this to the metrics infrastructure, but > > it'll likely take me some more time to get to it and to figure out (once > > again) how the messages versioning work. > > > > I think it is useful info to report to the MDS, but it's not required to > send these to the MDS to solve the current problem. My suggestion would > be to add what's needed to track these stats in the kclient and report > them via debugfs, but don't send the info to the MDS just yet. > > Later, we could extend the protocol with COPY stats, and add the > necessary infrastructure to the MDS to deal with it. Once that's in > place, we can then extend the kclient to start sending this info along > when it reports the stats. Awesome, that sounds good to me. I'll look into re-writing this patch following your suggestion. Thanks! Cheers, -- Luís
On 10/22/21 1:30 AM, Patrick Donnelly wrote: > On Thu, Oct 21, 2021 at 12:35 PM Jeff Layton <jlayton@kernel.org> wrote: >> On Thu, 2021-10-21 at 12:18 -0400, Patrick Donnelly wrote: >>> On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: >>>> On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: >>>>> On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: >>>>>> On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: >>>>>>> This counter will keep track of the number of remote object copies done on >>>>>>> copy_file_range syscalls. This counter will be filesystem per-client, and >>>>>>> can be accessed from the client debugfs directory. >>>>>>> >>>>>>> Cc: Patrick Donnelly <pdonnell@redhat.com> >>>>>>> Signed-off-by: Luís Henriques <lhenriques@suse.de> >>>>>>> --- >>>>>>> This is an RFC to reply to Patrick's request in [0]. Note that I'm not >>>>>>> 100% sure about the usefulness of this patch, or if this is the best way >>>>>>> to provide the functionality Patrick requested. Anyway, this is just to >>>>>>> get some feedback, hence the RFC. >>>>>>> >>>>>>> Cheers, >>>>>>> -- >>>>>>> Luís >>>>>>> >>>>>>> [0] https://github.com/ceph/ceph/pull/42720 >>>>>>> >>>>>> I think this would be better integrated into the stats infrastructure. >>>>>> >>>>>> Maybe you could add a new set of "copy" stats to struct >>>>>> ceph_client_metric that tracks the total copy operations done, their >>>>>> size and latency (similar to read and write ops)? >>>>> I think it's a good idea to integrate this into "stats" but I think a >>>>> local debugfs file for some counters is still useful. The "stats" >>>>> module is immature at this time and I'd rather not build any qa tests >>>>> (yet) that rely on it. >>>>> >>>>> Can we generalize this patch-set to a file named "op_counters" or >>>>> similar and additionally add other OSD ops performed by the kclient? >>>>> >>>> >>>> Tracking this sort of thing is the main purpose of the stats code. I'm >>>> really not keen on adding a whole separate set of files for reporting >>>> this. >>> Maybe I'm confused. Is there some "file" which is already used for >>> this type of debugging information? Or do you mean the code for >>> sending stats to the MDS to support cephfs-top? >>> >>>> What's the specific problem with relying on the data in debugfs >>>> "metrics" file? >>> Maybe no problem? I wasn't aware of a "metrics" file. >>> >> Yes. For instance: >> >> # cat /sys/kernel/debug/ceph/*/metrics >> item total >> ------------------------------------------ >> opened files / total inodes 0 / 4 >> pinned i_caps / total inodes 5 / 4 >> opened inodes / total inodes 0 / 4 >> >> item total avg_lat(us) min_lat(us) max_lat(us) stdev(us) >> ----------------------------------------------------------------------------------- >> read 0 0 0 0 0 >> write 5 914013 824797 1092343 103476 >> metadata 79 12856 1572 114572 13262 >> >> item total avg_sz(bytes) min_sz(bytes) max_sz(bytes) total_sz(bytes) >> ---------------------------------------------------------------------------------------- >> read 0 0 0 0 0 >> write 5 4194304 4194304 4194304 20971520 >> >> item total miss hit >> ------------------------------------------------- >> d_lease 11 0 29 >> caps 5 68 10702 >> >> >> I'm proposing that Luis add new lines for "copy" to go along with the >> "read" and "write" ones. The "total" counter should give you a count of >> the number of operations. > Okay that makes more sense! > > Side note: I am a bit horrified by how computer-unfriendly that > table-formatted data is. Any suggestion to improve this ? How about just make the "metric" file writable like a switch ? And as default it will show the data as above and if tools want the computer-friendly format, just write none-zero to it, then show raw data just like: # cat /sys/kernel/debug/ceph/*/metrics opened_files:0 pinned_i_caps:5 opened_inodes:0 total_inodes:4 read_latency:0,0,0,0,0 write_latency:5,914013,824797,1092343,103476 metadata_latency:79,12856,1572,114572,13262 read_size:0,0,0,0,0 write_size:5,4194304,4194304,4194304,20971520 d_lease:11,0,29 caps:5,68,10702
On Tue, 2021-10-26 at 11:05 +0800, Xiubo Li wrote: > On 10/22/21 1:30 AM, Patrick Donnelly wrote: > > On Thu, Oct 21, 2021 at 12:35 PM Jeff Layton <jlayton@kernel.org> wrote: > > > On Thu, 2021-10-21 at 12:18 -0400, Patrick Donnelly wrote: > > > > On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: > > > > > On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: > > > > > > On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > > > > > > > This counter will keep track of the number of remote object copies done on > > > > > > > > copy_file_range syscalls. This counter will be filesystem per-client, and > > > > > > > > can be accessed from the client debugfs directory. > > > > > > > > > > > > > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > > > > > > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > > > > > > > --- > > > > > > > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > > > > > > > 100% sure about the usefulness of this patch, or if this is the best way > > > > > > > > to provide the functionality Patrick requested. Anyway, this is just to > > > > > > > > get some feedback, hence the RFC. > > > > > > > > > > > > > > > > Cheers, > > > > > > > > -- > > > > > > > > Luís > > > > > > > > > > > > > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > > > > > > > > > > > > I think this would be better integrated into the stats infrastructure. > > > > > > > > > > > > > > Maybe you could add a new set of "copy" stats to struct > > > > > > > ceph_client_metric that tracks the total copy operations done, their > > > > > > > size and latency (similar to read and write ops)? > > > > > > I think it's a good idea to integrate this into "stats" but I think a > > > > > > local debugfs file for some counters is still useful. The "stats" > > > > > > module is immature at this time and I'd rather not build any qa tests > > > > > > (yet) that rely on it. > > > > > > > > > > > > Can we generalize this patch-set to a file named "op_counters" or > > > > > > similar and additionally add other OSD ops performed by the kclient? > > > > > > > > > > > > > > > > Tracking this sort of thing is the main purpose of the stats code. I'm > > > > > really not keen on adding a whole separate set of files for reporting > > > > > this. > > > > Maybe I'm confused. Is there some "file" which is already used for > > > > this type of debugging information? Or do you mean the code for > > > > sending stats to the MDS to support cephfs-top? > > > > > > > > > What's the specific problem with relying on the data in debugfs > > > > > "metrics" file? > > > > Maybe no problem? I wasn't aware of a "metrics" file. > > > > > > > Yes. For instance: > > > > > > # cat /sys/kernel/debug/ceph/*/metrics > > > item total > > > ------------------------------------------ > > > opened files / total inodes 0 / 4 > > > pinned i_caps / total inodes 5 / 4 > > > opened inodes / total inodes 0 / 4 > > > > > > item total avg_lat(us) min_lat(us) max_lat(us) stdev(us) > > > ----------------------------------------------------------------------------------- > > > read 0 0 0 0 0 > > > write 5 914013 824797 1092343 103476 > > > metadata 79 12856 1572 114572 13262 > > > > > > item total avg_sz(bytes) min_sz(bytes) max_sz(bytes) total_sz(bytes) > > > ---------------------------------------------------------------------------------------- > > > read 0 0 0 0 0 > > > write 5 4194304 4194304 4194304 20971520 > > > > > > item total miss hit > > > ------------------------------------------------- > > > d_lease 11 0 29 > > > caps 5 68 10702 > > > > > > > > > I'm proposing that Luis add new lines for "copy" to go along with the > > > "read" and "write" ones. The "total" counter should give you a count of > > > the number of operations. > > Okay that makes more sense! > > > > Side note: I am a bit horrified by how computer-unfriendly that > > table-formatted data is. > > Any suggestion to improve this ? > > How about just make the "metric" file writable like a switch ? And as > default it will show the data as above and if tools want the > computer-friendly format, just write none-zero to it, then show raw data > just like: > > # cat /sys/kernel/debug/ceph/*/metrics > opened_files:0 > pinned_i_caps:5 > opened_inodes:0 > total_inodes:4 > > read_latency:0,0,0,0,0 > write_latency:5,914013,824797,1092343,103476 > metadata_latency:79,12856,1572,114572,13262 > > read_size:0,0,0,0,0 > write_size:5,4194304,4194304,4194304,20971520 > > d_lease:11,0,29 > caps:5,68,10702 > > I'd rather not multiplex the output of this file based on some input. That would also be rather hard to do -- write() and read() are two different syscalls, so you'd need to track a bool (or something) across them somehow. Currently, I doubt there are many scripts in the field that scrape this info and debugfs is specifically excluded from ABI concerns. If we want to make it more machine-readable (which sounds like a good thing), then I suggest we just change the output to something like what you have above and not worry about preserving the "legacy" output.
On Tue, Oct 26, 2021 at 07:40:51AM -0400, Jeff Layton wrote: > On Tue, 2021-10-26 at 11:05 +0800, Xiubo Li wrote: > > On 10/22/21 1:30 AM, Patrick Donnelly wrote: > > > On Thu, Oct 21, 2021 at 12:35 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > On Thu, 2021-10-21 at 12:18 -0400, Patrick Donnelly wrote: > > > > > On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: > > > > > > > On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: > > > > > > > > On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: > > > > > > > > > This counter will keep track of the number of remote object copies done on > > > > > > > > > copy_file_range syscalls. This counter will be filesystem per-client, and > > > > > > > > > can be accessed from the client debugfs directory. > > > > > > > > > > > > > > > > > > Cc: Patrick Donnelly <pdonnell@redhat.com> > > > > > > > > > Signed-off-by: Luís Henriques <lhenriques@suse.de> > > > > > > > > > --- > > > > > > > > > This is an RFC to reply to Patrick's request in [0]. Note that I'm not > > > > > > > > > 100% sure about the usefulness of this patch, or if this is the best way > > > > > > > > > to provide the functionality Patrick requested. Anyway, this is just to > > > > > > > > > get some feedback, hence the RFC. > > > > > > > > > > > > > > > > > > Cheers, > > > > > > > > > -- > > > > > > > > > Luís > > > > > > > > > > > > > > > > > > [0] https://github.com/ceph/ceph/pull/42720 > > > > > > > > > > > > > > > > > I think this would be better integrated into the stats infrastructure. > > > > > > > > > > > > > > > > Maybe you could add a new set of "copy" stats to struct > > > > > > > > ceph_client_metric that tracks the total copy operations done, their > > > > > > > > size and latency (similar to read and write ops)? > > > > > > > I think it's a good idea to integrate this into "stats" but I think a > > > > > > > local debugfs file for some counters is still useful. The "stats" > > > > > > > module is immature at this time and I'd rather not build any qa tests > > > > > > > (yet) that rely on it. > > > > > > > > > > > > > > Can we generalize this patch-set to a file named "op_counters" or > > > > > > > similar and additionally add other OSD ops performed by the kclient? > > > > > > > > > > > > > > > > > > > Tracking this sort of thing is the main purpose of the stats code. I'm > > > > > > really not keen on adding a whole separate set of files for reporting > > > > > > this. > > > > > Maybe I'm confused. Is there some "file" which is already used for > > > > > this type of debugging information? Or do you mean the code for > > > > > sending stats to the MDS to support cephfs-top? > > > > > > > > > > > What's the specific problem with relying on the data in debugfs > > > > > > "metrics" file? > > > > > Maybe no problem? I wasn't aware of a "metrics" file. > > > > > > > > > Yes. For instance: > > > > > > > > # cat /sys/kernel/debug/ceph/*/metrics > > > > item total > > > > ------------------------------------------ > > > > opened files / total inodes 0 / 4 > > > > pinned i_caps / total inodes 5 / 4 > > > > opened inodes / total inodes 0 / 4 > > > > > > > > item total avg_lat(us) min_lat(us) max_lat(us) stdev(us) > > > > ----------------------------------------------------------------------------------- > > > > read 0 0 0 0 0 > > > > write 5 914013 824797 1092343 103476 > > > > metadata 79 12856 1572 114572 13262 > > > > > > > > item total avg_sz(bytes) min_sz(bytes) max_sz(bytes) total_sz(bytes) > > > > ---------------------------------------------------------------------------------------- > > > > read 0 0 0 0 0 > > > > write 5 4194304 4194304 4194304 20971520 > > > > > > > > item total miss hit > > > > ------------------------------------------------- > > > > d_lease 11 0 29 > > > > caps 5 68 10702 > > > > > > > > > > > > I'm proposing that Luis add new lines for "copy" to go along with the > > > > "read" and "write" ones. The "total" counter should give you a count of > > > > the number of operations. > > > Okay that makes more sense! > > > > > > Side note: I am a bit horrified by how computer-unfriendly that > > > table-formatted data is. > > > > Any suggestion to improve this ? > > > > How about just make the "metric" file writable like a switch ? And as > > default it will show the data as above and if tools want the > > computer-friendly format, just write none-zero to it, then show raw data > > just like: > > > > # cat /sys/kernel/debug/ceph/*/metrics > > opened_files:0 > > pinned_i_caps:5 > > opened_inodes:0 > > total_inodes:4 > > > > read_latency:0,0,0,0,0 > > write_latency:5,914013,824797,1092343,103476 > > metadata_latency:79,12856,1572,114572,13262 > > > > read_size:0,0,0,0,0 > > write_size:5,4194304,4194304,4194304,20971520 > > > > d_lease:11,0,29 > > caps:5,68,10702 > > > > > > I'd rather not multiplex the output of this file based on some input. > That would also be rather hard to do -- write() and read() are two > different syscalls, so you'd need to track a bool (or something) across > them somehow. > > Currently, I doubt there are many scripts in the field that scrape this > info and debugfs is specifically excluded from ABI concerns. If we want > to make it more machine-readable (which sounds like a good thing), then > I suggest we just change the output to something like what you have > above and not worry about preserving the "legacy" output. Ok, before submitting any new revision of this patch I should probably clean this up. I can submit a patch to change the format to what Xiubo is proposing. Obviously, that patch will also need to document what all those fields actually mean. Alternatively, the metrics file could be changed into a directory and have 4 different files, one per each section: metrics/ |- files <-- not sure how to name the 1st section |- latency |- size \- caps Each of these files would then have the header but, since it's a single header, parsing it in a script would be pretty easy. The advantage is that this would be self-documented (with filenames and headers). Cheers, -- Luís
On 10/26/21 7:40 PM, Jeff Layton wrote: > On Tue, 2021-10-26 at 11:05 +0800, Xiubo Li wrote: >> On 10/22/21 1:30 AM, Patrick Donnelly wrote: >>> On Thu, Oct 21, 2021 at 12:35 PM Jeff Layton <jlayton@kernel.org> wrote: >>>> On Thu, 2021-10-21 at 12:18 -0400, Patrick Donnelly wrote: >>>>> On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: >>>>>> On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: >>>>>>> On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: >>>>>>>> On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: >>>>>>>>> This counter will keep track of the number of remote object copies done on >>>>>>>>> copy_file_range syscalls. This counter will be filesystem per-client, and >>>>>>>>> can be accessed from the client debugfs directory. >>>>>>>>> >>>>>>>>> Cc: Patrick Donnelly <pdonnell@redhat.com> >>>>>>>>> Signed-off-by: Luís Henriques <lhenriques@suse.de> >>>>>>>>> --- >>>>>>>>> This is an RFC to reply to Patrick's request in [0]. Note that I'm not >>>>>>>>> 100% sure about the usefulness of this patch, or if this is the best way >>>>>>>>> to provide the functionality Patrick requested. Anyway, this is just to >>>>>>>>> get some feedback, hence the RFC. >>>>>>>>> >>>>>>>>> Cheers, >>>>>>>>> -- >>>>>>>>> Luís >>>>>>>>> >>>>>>>>> [0] https://github.com/ceph/ceph/pull/42720 >>>>>>>>> >>>>>>>> I think this would be better integrated into the stats infrastructure. >>>>>>>> >>>>>>>> Maybe you could add a new set of "copy" stats to struct >>>>>>>> ceph_client_metric that tracks the total copy operations done, their >>>>>>>> size and latency (similar to read and write ops)? >>>>>>> I think it's a good idea to integrate this into "stats" but I think a >>>>>>> local debugfs file for some counters is still useful. The "stats" >>>>>>> module is immature at this time and I'd rather not build any qa tests >>>>>>> (yet) that rely on it. >>>>>>> >>>>>>> Can we generalize this patch-set to a file named "op_counters" or >>>>>>> similar and additionally add other OSD ops performed by the kclient? >>>>>>> >>>>>> Tracking this sort of thing is the main purpose of the stats code. I'm >>>>>> really not keen on adding a whole separate set of files for reporting >>>>>> this. >>>>> Maybe I'm confused. Is there some "file" which is already used for >>>>> this type of debugging information? Or do you mean the code for >>>>> sending stats to the MDS to support cephfs-top? >>>>> >>>>>> What's the specific problem with relying on the data in debugfs >>>>>> "metrics" file? >>>>> Maybe no problem? I wasn't aware of a "metrics" file. >>>>> >>>> Yes. For instance: >>>> >>>> # cat /sys/kernel/debug/ceph/*/metrics >>>> item total >>>> ------------------------------------------ >>>> opened files / total inodes 0 / 4 >>>> pinned i_caps / total inodes 5 / 4 >>>> opened inodes / total inodes 0 / 4 >>>> >>>> item total avg_lat(us) min_lat(us) max_lat(us) stdev(us) >>>> ----------------------------------------------------------------------------------- >>>> read 0 0 0 0 0 >>>> write 5 914013 824797 1092343 103476 >>>> metadata 79 12856 1572 114572 13262 >>>> >>>> item total avg_sz(bytes) min_sz(bytes) max_sz(bytes) total_sz(bytes) >>>> ---------------------------------------------------------------------------------------- >>>> read 0 0 0 0 0 >>>> write 5 4194304 4194304 4194304 20971520 >>>> >>>> item total miss hit >>>> ------------------------------------------------- >>>> d_lease 11 0 29 >>>> caps 5 68 10702 >>>> >>>> >>>> I'm proposing that Luis add new lines for "copy" to go along with the >>>> "read" and "write" ones. The "total" counter should give you a count of >>>> the number of operations. >>> Okay that makes more sense! >>> >>> Side note: I am a bit horrified by how computer-unfriendly that >>> table-formatted data is. >> Any suggestion to improve this ? >> >> How about just make the "metric" file writable like a switch ? And as >> default it will show the data as above and if tools want the >> computer-friendly format, just write none-zero to it, then show raw data >> just like: >> >> # cat /sys/kernel/debug/ceph/*/metrics >> opened_files:0 >> pinned_i_caps:5 >> opened_inodes:0 >> total_inodes:4 >> >> read_latency:0,0,0,0,0 >> write_latency:5,914013,824797,1092343,103476 >> metadata_latency:79,12856,1572,114572,13262 >> >> read_size:0,0,0,0,0 >> write_size:5,4194304,4194304,4194304,20971520 >> >> d_lease:11,0,29 >> caps:5,68,10702 >> >> > I'd rather not multiplex the output of this file based on some input. > That would also be rather hard to do -- write() and read() are two > different syscalls, so you'd need to track a bool (or something) across > them somehow. > > Currently, I doubt there are many scripts in the field that scrape this > info and debugfs is specifically excluded from ABI concerns. If we want > to make it more machine-readable (which sounds like a good thing), then > I suggest we just change the output to something like what you have > above and not worry about preserving the "legacy" output. Sound good to me.
On 10/26/21 11:31 PM, Luís Henriques wrote: > On Tue, Oct 26, 2021 at 07:40:51AM -0400, Jeff Layton wrote: >> On Tue, 2021-10-26 at 11:05 +0800, Xiubo Li wrote: >>> On 10/22/21 1:30 AM, Patrick Donnelly wrote: >>>> On Thu, Oct 21, 2021 at 12:35 PM Jeff Layton <jlayton@kernel.org> wrote: >>>>> On Thu, 2021-10-21 at 12:18 -0400, Patrick Donnelly wrote: >>>>>> On Thu, Oct 21, 2021 at 11:44 AM Jeff Layton <jlayton@kernel.org> wrote: >>>>>>> On Thu, 2021-10-21 at 09:52 -0400, Patrick Donnelly wrote: >>>>>>>> On Wed, Oct 20, 2021 at 12:27 PM Jeff Layton <jlayton@kernel.org> wrote: >>>>>>>>> On Wed, 2021-10-20 at 15:37 +0100, Luís Henriques wrote: >>>>>>>>>> This counter will keep track of the number of remote object copies done on >>>>>>>>>> copy_file_range syscalls. This counter will be filesystem per-client, and >>>>>>>>>> can be accessed from the client debugfs directory. >>>>>>>>>> >>>>>>>>>> Cc: Patrick Donnelly <pdonnell@redhat.com> >>>>>>>>>> Signed-off-by: Luís Henriques <lhenriques@suse.de> >>>>>>>>>> --- >>>>>>>>>> This is an RFC to reply to Patrick's request in [0]. Note that I'm not >>>>>>>>>> 100% sure about the usefulness of this patch, or if this is the best way >>>>>>>>>> to provide the functionality Patrick requested. Anyway, this is just to >>>>>>>>>> get some feedback, hence the RFC. >>>>>>>>>> >>>>>>>>>> Cheers, >>>>>>>>>> -- >>>>>>>>>> Luís >>>>>>>>>> >>>>>>>>>> [0] https://github.com/ceph/ceph/pull/42720 >>>>>>>>>> >>>>>>>>> I think this would be better integrated into the stats infrastructure. >>>>>>>>> >>>>>>>>> Maybe you could add a new set of "copy" stats to struct >>>>>>>>> ceph_client_metric that tracks the total copy operations done, their >>>>>>>>> size and latency (similar to read and write ops)? >>>>>>>> I think it's a good idea to integrate this into "stats" but I think a >>>>>>>> local debugfs file for some counters is still useful. The "stats" >>>>>>>> module is immature at this time and I'd rather not build any qa tests >>>>>>>> (yet) that rely on it. >>>>>>>> >>>>>>>> Can we generalize this patch-set to a file named "op_counters" or >>>>>>>> similar and additionally add other OSD ops performed by the kclient? >>>>>>>> >>>>>>> Tracking this sort of thing is the main purpose of the stats code. I'm >>>>>>> really not keen on adding a whole separate set of files for reporting >>>>>>> this. >>>>>> Maybe I'm confused. Is there some "file" which is already used for >>>>>> this type of debugging information? Or do you mean the code for >>>>>> sending stats to the MDS to support cephfs-top? >>>>>> >>>>>>> What's the specific problem with relying on the data in debugfs >>>>>>> "metrics" file? >>>>>> Maybe no problem? I wasn't aware of a "metrics" file. >>>>>> >>>>> Yes. For instance: >>>>> >>>>> # cat /sys/kernel/debug/ceph/*/metrics >>>>> item total >>>>> ------------------------------------------ >>>>> opened files / total inodes 0 / 4 >>>>> pinned i_caps / total inodes 5 / 4 >>>>> opened inodes / total inodes 0 / 4 >>>>> >>>>> item total avg_lat(us) min_lat(us) max_lat(us) stdev(us) >>>>> ----------------------------------------------------------------------------------- >>>>> read 0 0 0 0 0 >>>>> write 5 914013 824797 1092343 103476 >>>>> metadata 79 12856 1572 114572 13262 >>>>> >>>>> item total avg_sz(bytes) min_sz(bytes) max_sz(bytes) total_sz(bytes) >>>>> ---------------------------------------------------------------------------------------- >>>>> read 0 0 0 0 0 >>>>> write 5 4194304 4194304 4194304 20971520 >>>>> >>>>> item total miss hit >>>>> ------------------------------------------------- >>>>> d_lease 11 0 29 >>>>> caps 5 68 10702 >>>>> >>>>> >>>>> I'm proposing that Luis add new lines for "copy" to go along with the >>>>> "read" and "write" ones. The "total" counter should give you a count of >>>>> the number of operations. >>>> Okay that makes more sense! >>>> >>>> Side note: I am a bit horrified by how computer-unfriendly that >>>> table-formatted data is. >>> Any suggestion to improve this ? >>> >>> How about just make the "metric" file writable like a switch ? And as >>> default it will show the data as above and if tools want the >>> computer-friendly format, just write none-zero to it, then show raw data >>> just like: >>> >>> # cat /sys/kernel/debug/ceph/*/metrics >>> opened_files:0 >>> pinned_i_caps:5 >>> opened_inodes:0 >>> total_inodes:4 >>> >>> read_latency:0,0,0,0,0 >>> write_latency:5,914013,824797,1092343,103476 >>> metadata_latency:79,12856,1572,114572,13262 >>> >>> read_size:0,0,0,0,0 >>> write_size:5,4194304,4194304,4194304,20971520 >>> >>> d_lease:11,0,29 >>> caps:5,68,10702 >>> >>> >> I'd rather not multiplex the output of this file based on some input. >> That would also be rather hard to do -- write() and read() are two >> different syscalls, so you'd need to track a bool (or something) across >> them somehow. >> >> Currently, I doubt there are many scripts in the field that scrape this >> info and debugfs is specifically excluded from ABI concerns. If we want >> to make it more machine-readable (which sounds like a good thing), then >> I suggest we just change the output to something like what you have >> above and not worry about preserving the "legacy" output. > Ok, before submitting any new revision of this patch I should probably > clean this up. I can submit a patch to change the format to what Xiubo is > proposing. Obviously, that patch will also need to document what all > those fields actually mean. > > Alternatively, the metrics file could be changed into a directory and have > 4 different files, one per each section: > > metrics/ > |- files <-- not sure how to name the 1st section > |- latency > |- size > \- caps > > Each of these files would then have the header but, since it's a single > header, parsing it in a script would be pretty easy. The advantage is > that this would be self-documented (with filenames and headers). This sounds good to me. > > Cheers, > -- > Luís >
diff --git a/fs/ceph/debugfs.c b/fs/ceph/debugfs.c index 38b78b45811f..09f4c04ade0e 100644 --- a/fs/ceph/debugfs.c +++ b/fs/ceph/debugfs.c @@ -346,13 +346,22 @@ static int status_show(struct seq_file *s, void *p) return 0; } +static int copyfrom_show(struct seq_file *s, void *p) +{ + struct ceph_fs_client *fsc = s->private; + + seq_printf(s, "%llu\n", atomic64_read(&fsc->copyfrom_count)); + + return 0; +} + DEFINE_SHOW_ATTRIBUTE(mdsmap); DEFINE_SHOW_ATTRIBUTE(mdsc); DEFINE_SHOW_ATTRIBUTE(caps); DEFINE_SHOW_ATTRIBUTE(mds_sessions); DEFINE_SHOW_ATTRIBUTE(metric); DEFINE_SHOW_ATTRIBUTE(status); - +DEFINE_SHOW_ATTRIBUTE(copyfrom); /* * debugfs @@ -387,6 +396,7 @@ void ceph_fs_debugfs_cleanup(struct ceph_fs_client *fsc) debugfs_remove(fsc->debugfs_caps); debugfs_remove(fsc->debugfs_metric); debugfs_remove(fsc->debugfs_mdsc); + debugfs_remove(fsc->debugfs_copyfrom); } void ceph_fs_debugfs_init(struct ceph_fs_client *fsc) @@ -443,6 +453,11 @@ void ceph_fs_debugfs_init(struct ceph_fs_client *fsc) fsc->client->debugfs_dir, fsc, &status_fops); + fsc->debugfs_copyfrom = debugfs_create_file("copyfrom", + 0400, + fsc->client->debugfs_dir, + fsc, + ©from_fops); } diff --git a/fs/ceph/file.c b/fs/ceph/file.c index d16fd2d5fd42..bbeb437ca4bf 100644 --- a/fs/ceph/file.c +++ b/fs/ceph/file.c @@ -2254,6 +2254,7 @@ static ssize_t ceph_do_objects_copy(struct ceph_inode_info *src_ci, u64 *src_off bytes = ret; goto out; } + atomic64_inc(&fsc->copyfrom_count); len -= object_size; bytes += object_size; *src_off += object_size; diff --git a/fs/ceph/super.c b/fs/ceph/super.c index 9b1b7f4cfdd4..4972554185e3 100644 --- a/fs/ceph/super.c +++ b/fs/ceph/super.c @@ -670,6 +670,7 @@ static struct ceph_fs_client *create_fs_client(struct ceph_mount_options *fsopt, fsc->have_copy_from2 = true; atomic_long_set(&fsc->writeback_count, 0); + atomic64_set(&fsc->copyfrom_count, 0); err = -ENOMEM; /* diff --git a/fs/ceph/super.h b/fs/ceph/super.h index a40eb14c282a..65846beca418 100644 --- a/fs/ceph/super.h +++ b/fs/ceph/super.h @@ -119,6 +119,7 @@ struct ceph_fs_client { struct ceph_mds_client *mdsc; atomic_long_t writeback_count; + atomic64_t copyfrom_count; struct workqueue_struct *inode_wq; struct workqueue_struct *cap_wq; @@ -131,6 +132,7 @@ struct ceph_fs_client { struct dentry *debugfs_metric; struct dentry *debugfs_status; struct dentry *debugfs_mds_sessions; + struct dentry *debugfs_copyfrom; #endif #ifdef CONFIG_CEPH_FSCACHE
This counter will keep track of the number of remote object copies done on copy_file_range syscalls. This counter will be filesystem per-client, and can be accessed from the client debugfs directory. Cc: Patrick Donnelly <pdonnell@redhat.com> Signed-off-by: Luís Henriques <lhenriques@suse.de> --- This is an RFC to reply to Patrick's request in [0]. Note that I'm not 100% sure about the usefulness of this patch, or if this is the best way to provide the functionality Patrick requested. Anyway, this is just to get some feedback, hence the RFC. Cheers, -- Luís [0] https://github.com/ceph/ceph/pull/42720 fs/ceph/debugfs.c | 17 ++++++++++++++++- fs/ceph/file.c | 1 + fs/ceph/super.c | 1 + fs/ceph/super.h | 2 ++ 4 files changed, 20 insertions(+), 1 deletion(-)