Ceph

h1. 1H - Inline data support

h3. Live Pad

The live pad can be found here: "[pad]":http://pad.ceph.com/p/Inline_data_support_for_Ceph

h3. Summit Snapshot

Coding tasks

# Insert extended attribute of file for MDS to store data of small files

## piggyback on the File cap bits, existing cap writeback mechanism

# MDS uninline file content when it goes into the MIX state

# add fields to MClientCaps

# Inline data is returned to the client via encode_inodestat() (used by lookup, readdir, stat, open, etc.)

# MDS would store inline data inside inode_t (bufferlist inline_data)

# Client (libcephfs, ceph-fuse)

# if size is small and we are flushing, flush inline to mds

## prototype and refine protocol changes

# Linux kernel client

## read side

### copy into page cache from inode buffer from readpage()

## write side

### writepage[s]() ..

### begin_page_writeback() ???? somethign like that... set the writeback bit, lock page

### if (size is small and we want to inline) {

### copy into the inode buffer

### trigger mds cap flush

### wait for flush

### } else {

### do the regular thing

### }

### end_page_writeback()

Documentation tasks

# Document the communication protocol

1 Client side

1.1 ceph_write_end()

<pre>

          if (inode->status == INLINED) {

                 if (write_pos < PAGE_SIZE) {

                     write_page_to_inode();

                     err = mark_inode_dirty();

                     if (err == ESTATUS) // status has changed to NOTINLINING or NOTINLINED

                         write_page_to_osd();

                     return;

                 }

                 if (write_pos > PAGE_SIZE) {

                    inode->status = NOTINLINING;

                    mark_inode_dirty(); // ansynchoronously tell mds to change status to NOTINLINING

                 } 

                 if (the interval [write_pos, write_pos + write_len] overlap with the interval [0, PAGE_SIZE]) {

                     inode->status = NOTINLINED;

                     mark_inode_dirty();

                 }

          }

          if (inode->status == NOTINLINING) {

                 if (the interval [write_pos, write_pos + write_len] overlap with the interval [0, PAGE_SIZE]) {

                     inode->status = NOTINLINED;

                     mark_inode_dirty();

                 }

          }

</pre>

1.2 write_page()

<pre>

          if (page->index == 0 && inode->status == INLINED) { // for mmap(), it won't go through write_end

                err = write_page_to_inode();

                mark_inode_dirty();

          }

          write_page_to_osd();

</pre>

1.3  read_page()

<pre>

          if (page->index == 0 && (inode->status == INLINED || inode->status == NOTINLING)) {

                   err = copy_data_from_inode();

                   if (err == ESTATUS) // status has changed to NOTINLINED

                          read_page_from_osd();

                   return;

           }

           read_page_from_osd();

</pre>