Ceph » CephFS

We often see:

2018-07-03 03:10:20.042188 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x10014248370, dropping
2018-07-03 03:10:20.042191 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x1001424a92f, dropping
2018-07-03 03:10:20.042195 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x100142088f0, dropping
2018-07-03 03:10:20.042199 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x1001424bd21, dropping
2018-07-03 03:10:20.042201 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x10014250aa7, dropping
2018-07-03 03:10:20.042205 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x1001420892b, dropping
2018-07-03 03:10:20.042207 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x1001424804b, dropping
2018-07-03 03:10:20.042211 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x10014253b43, dropping
2018-07-03 03:10:20.042215 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x10014247786, dropping
2018-07-03 03:10:20.042221 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x10014246826, dropping
2018-07-03 03:10:20.042225 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x1001424908f, dropping
2018-07-03 03:10:20.042228 7fc717bf9700  1 mds.0.locker
handle_client_caps on unknown ino 0x100142496b9, dropping

as a normal consequence of subtree migration. From Zheng:

"handle_client_caps on unknown ino xxx” is normal after migrating a subtree

Sequence of events that cause this are:

- client sends caps message to mds.a
- mds finishes subtree migration, send cap export to client
- mds trim its cache
- mds receives cap messages from client

Raise debugging of this message so it doesn't flood the log.