655 lines
22 KiB
Plaintext
655 lines
22 KiB
Plaintext
What: /sys/fs/lustre/version
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows current running lustre version.
|
|
|
|
What: /sys/fs/lustre/pinger
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows if the lustre module has pinger support.
|
|
"on" means yes and "off" means no.
|
|
|
|
What: /sys/fs/lustre/health_check
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows whenever current system state believed to be "healthy",
|
|
"NOT HEALTHY", or "LBUG" whenever lustre has experienced
|
|
an internal assertion failure
|
|
|
|
What: /sys/fs/lustre/jobid_name
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Currently running job "name" for this node to be transferred
|
|
to Lustre servers for purposes of QoS and statistics gathering.
|
|
Writing into this file will change the name, reading outputs
|
|
currently set value.
|
|
|
|
What: /sys/fs/lustre/jobid_var
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Control file for lustre "jobstats" functionality, write new
|
|
value from the list below to change the mode:
|
|
disable - disable job name reporting to the servers (default)
|
|
procname_uid - form the job name as the current running
|
|
command name and pid with a dot in between
|
|
e.g. dd.1253
|
|
nodelocal - use jobid_name value from above.
|
|
|
|
What: /sys/fs/lustre/timeout
|
|
Date: June 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls "lustre timeout" variable, also known as obd_timeout
|
|
in some old manual. In the past obd_timeout was of paramount
|
|
importance as the timeout value used everywhere and where
|
|
other timeouts were derived from. These days it's much less
|
|
important as network timeouts are mostly determined by
|
|
AT (adaptive timeouts).
|
|
Unit: seconds, default: 100
|
|
|
|
What: /sys/fs/lustre/max_dirty_mb
|
|
Date: June 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls total number of dirty cache (in megabytes) allowed
|
|
across all mounted lustre filesystems.
|
|
Since writeout of dirty pages in Lustre is somewhat expensive,
|
|
when you allow to many dirty pages, this might lead to
|
|
performance degradations as kernel tries to desperately
|
|
find some pages to free/writeout.
|
|
Default 1/2 RAM. Min value 4, max value 9/10 of RAM.
|
|
|
|
What: /sys/fs/lustre/debug_peer_on_timeout
|
|
Date: June 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Control if lnet debug information should be printed when
|
|
an RPC timeout occurs.
|
|
0 disabled (default)
|
|
1 enabled
|
|
|
|
What: /sys/fs/lustre/dump_on_timeout
|
|
Date: June 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls if Lustre debug log should be dumped when an RPC
|
|
timeout occurs. This is useful if yout debug buffer typically
|
|
rolls over by the time you notice RPC timeouts.
|
|
|
|
What: /sys/fs/lustre/dump_on_eviction
|
|
Date: June 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls if Lustre debug log should be dumped when an this
|
|
client is evicted from one of the servers.
|
|
This is useful if yout debug buffer typically rolls over
|
|
by the time you notice the eviction event.
|
|
|
|
What: /sys/fs/lustre/at_min
|
|
Date: July 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls minimum adaptive timeout in seconds. If you encounter
|
|
a case where clients timeout due to server-reported processing
|
|
time being too short, you might consider increasing this value.
|
|
One common case of this if the underlying network has
|
|
unpredictable long delays.
|
|
Default: 0
|
|
|
|
What: /sys/fs/lustre/at_max
|
|
Date: July 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls maximum adaptive timeout in seconds. If at_max timeout
|
|
is reached for an RPC, the RPC will time out.
|
|
Some genuinuely slow network hardware might warrant increasing
|
|
this value.
|
|
Setting this value to 0 disables Adaptive Timeouts
|
|
functionality and old-style obd_timeout value is then used.
|
|
Default: 600
|
|
|
|
What: /sys/fs/lustre/at_extra
|
|
Date: July 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls how much extra time to request for unfinished requests
|
|
in processing in seconds. Normally a server-side parameter, it
|
|
is also used on the client for responses to various LDLM ASTs
|
|
that are handled with a special server thread on the client.
|
|
This is a way for the servers to ask the clients not to time
|
|
out the request that reached current servicing time estimate
|
|
yet and give it some more time.
|
|
Default: 30
|
|
|
|
What: /sys/fs/lustre/at_early_margin
|
|
Date: July 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls when to send the early reply for requests that are
|
|
about to timeout as an offset to the estimated service time in
|
|
seconds..
|
|
Default: 5
|
|
|
|
What: /sys/fs/lustre/at_history
|
|
Date: July 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls for how many seconds to remember slowest events
|
|
encountered by adaptive timeouts code.
|
|
Default: 600
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/blocksize
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Biggest blocksize on object storage server for this filesystem.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/kbytestotal
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows total number of kilobytes of space on this filesystem
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/kbytesfree
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows total number of free kilobytes of space on this filesystem
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/kbytesavail
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows total number of free kilobytes of space on this filesystem
|
|
actually available for use (taking into account per-client
|
|
grants and filesystem reservations).
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/filestotal
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows total number of inodes on the filesystem.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/filesfree
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows estimated number of free inodes on the filesystem
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/client_type
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows whenever this filesystem considers this client to be
|
|
compute cluster-local or remote. Remote clients have
|
|
additional uid/gid convrting logic applied.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/fstype
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows filesystem type of the filesystem
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/uuid
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows this filesystem superblock uuid
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/max_read_ahead_mb
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Sets maximum number of megabytes in system memory to be
|
|
given to read-ahead cache.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/max_read_ahead_per_file_mb
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Sets maximum number of megabytes to read-ahead for a single file
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/max_read_ahead_whole_mb
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
For small reads, how many megabytes to actually request from
|
|
the server as initial read-ahead.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/checksum_pages
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Enables or disables per-page checksum at llite layer, before
|
|
the pages are actually given to lower level for network transfer
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/stats_track_pid
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Limit Lustre vfs operations gathering to just a single pid.
|
|
0 to track everything.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/stats_track_ppid
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Limit Lustre vfs operations gathering to just a single ppid.
|
|
0 to track everything.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/stats_track_gid
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Limit Lustre vfs operations gathering to just a single gid.
|
|
0 to track everything.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/statahead_max
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls maximum number of statahead requests to send when
|
|
sequential readdir+stat pattern is detected.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/statahead_agl
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls if AGL (async glimpse ahead - obtain object information
|
|
from OSTs in parallel with MDS during statahead) should be
|
|
enabled or disabled.
|
|
0 to disable, 1 to enable.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/lazystatfs
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls statfs(2) behaviour in the face of down servers.
|
|
If 0, always wait for all servers to come online,
|
|
if 1, ignote inactive servers.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/max_easize
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows maximum number of bytes file striping data could be
|
|
in current configuration of storage.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/default_easize
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows maximum observed file striping data seen by this
|
|
filesystem client instance.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/xattr_cache
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls extended attributes client-side cache.
|
|
1 to enable, 0 to disable.
|
|
|
|
What: /sys/fs/lustre/llite/<fsname>-<uuid>/unstable_stats
|
|
Date: Apr 2016
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows number of pages that were sent and acknowledged by
|
|
server but were not yet committed and therefore still
|
|
pinned in client memory even though no longer dirty.
|
|
|
|
What: /sys/fs/lustre/ldlm/cancel_unused_locks_before_replay
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls if client should replay unused locks during recovery
|
|
If a client tends to have a lot of unused locks in LRU,
|
|
recovery times might become prolonged.
|
|
1 - just locally cancel unused locks (default)
|
|
0 - replay unused locks.
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/resource_count
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Displays number of lock resources (objects on which individual
|
|
locks are taken) currently allocated in this namespace.
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/lock_count
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Displays number or locks allocated in this namespace.
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/lru_size
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls and displays LRU size limit for unused locks for this
|
|
namespace.
|
|
0 - LRU size is unlimited, controlled by server resources
|
|
positive number - number of locks to allow in lock LRU list
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/lock_unused_count
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Display number of locks currently sitting in the LRU list
|
|
of this namespace
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/lru_max_age
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Maximum number of milliseconds a lock could sit in LRU list
|
|
before client would voluntarily cancel it as unused.
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/early_lock_cancel
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls "early lock cancellation" feature on this namespace
|
|
if supported by the server.
|
|
When enabled, tries to preemtively cancel locks that would be
|
|
cancelled by verious operations and bundle the cancellation
|
|
requests in the same RPC as the main operation, which results
|
|
in significant speedups due to reduced lock-pingpong RPCs.
|
|
0 - disabled
|
|
1 - enabled (default)
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/pool/granted
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Displays number of granted locks in this namespace
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/pool/grant_rate
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of granted locks in this namespace during last
|
|
time interval
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/pool/cancel_rate
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of lock cancellations in this namespace during
|
|
last time interval
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/pool/grant_speed
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Calculated speed of lock granting (grant_rate - cancel_rate)
|
|
in this namespace
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/pool/grant_plan
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Estimated number of locks to be granted in the next time
|
|
interval in this namespace
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/pool/limit
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls number of allowed locks in this pool.
|
|
When lru_size is 0, this is the actual limit then.
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/pool/lock_volume_factor
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Multiplier for all lock volume calculations above.
|
|
Default is 1. Increase to make the client to more agressively
|
|
clean it's lock LRU list for this namespace.
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/pool/server_lock_volume
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Calculated server lock volume.
|
|
|
|
What: /sys/fs/lustre/ldlm/namespaces/<name>/pool/recalc_period
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls length of time between recalculation of above
|
|
values (in seconds).
|
|
|
|
What: /sys/fs/lustre/ldlm/services/ldlm_cbd/threads_min
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls minimum number of ldlm callback threads to start.
|
|
|
|
What: /sys/fs/lustre/ldlm/services/ldlm_cbd/threads_max
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls maximum number of ldlm callback threads to start.
|
|
|
|
What: /sys/fs/lustre/ldlm/services/ldlm_cbd/threads_started
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows actual number of ldlm callback threads running.
|
|
|
|
What: /sys/fs/lustre/ldlm/services/ldlm_cbd/high_priority_ratio
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls what percentage of ldlm callback threads is dedicated
|
|
to "high priority" incoming requests.
|
|
|
|
What: /sys/fs/lustre/{obdtype}/{connection_name}/blocksize
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Blocksize on backend filesystem for service behind this obd
|
|
device (or biggest blocksize for compound devices like lov
|
|
and lmv)
|
|
|
|
What: /sys/fs/lustre/{obdtype}/{connection_name}/kbytestotal
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Total number of kilobytes of space on backend filesystem
|
|
for service behind this obd (or total amount for compound
|
|
devices like lov lmv)
|
|
|
|
What: /sys/fs/lustre/{obdtype}/{connection_name}/kbytesfree
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of free kilobytes on backend filesystem for service
|
|
behind this obd (or total amount for compound devices
|
|
like lov lmv)
|
|
|
|
What: /sys/fs/lustre/{obdtype}/{connection_name}/kbytesavail
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of kilobytes of free space on backend filesystem
|
|
for service behind this obd (or total amount for compound
|
|
devices like lov lmv) that is actually available for use
|
|
(taking into account per-client and filesystem reservations).
|
|
|
|
What: /sys/fs/lustre/{obdtype}/{connection_name}/filestotal
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of inodes on backend filesystem for service behind this
|
|
obd.
|
|
|
|
What: /sys/fs/lustre/{obdtype}/{connection_name}/filesfree
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of free inodes on backend filesystem for service
|
|
behind this obd.
|
|
|
|
What: /sys/fs/lustre/mdc/{connection_name}/max_pages_per_rpc
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Maximum number of readdir pages to fit into a single readdir
|
|
RPC.
|
|
|
|
What: /sys/fs/lustre/{mdc,osc}/{connection_name}/max_rpcs_in_flight
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Maximum number of parallel RPCs on the wire to allow on
|
|
this connection. Increasing this number would help on higher
|
|
latency links, but has a chance of overloading a server
|
|
if you have too many clients like this.
|
|
Default: 8
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/max_pages_per_rpc
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Maximum number of pages to fit into a single RPC.
|
|
Typically bigger RPCs allow for better performance.
|
|
Default: however many pages to form 1M of data (256 pages
|
|
for 4K page sized platforms)
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/active
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls accessibility of this connection. If set to 0,
|
|
fail all accesses immediately.
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/checksums
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls whenever to checksum bulk RPC data over the wire
|
|
to this target.
|
|
1: enable (default) ; 0: disable
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/contention_seconds
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls for how long to consider a file contended once
|
|
indicated as such by the server.
|
|
When a file is considered contended, all operations switch to
|
|
synchronous lockless mode to avoid cache and lock pingpong.
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/cur_dirty_bytes
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Displays how many dirty bytes is presently in the cache for this
|
|
target.
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/cur_grant_bytes
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows how many bytes we have as a "dirty cache" grant from the
|
|
server. Writing a value smaller than shown allows to release
|
|
some grant back to the server.
|
|
Dirty cache grant is a way Lustre ensures that cached successful
|
|
writes on client do not end up discarded by the server due to
|
|
lack of space later on.
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/cur_lost_grant_bytes
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Shows how many granted bytes were released to the server due
|
|
to lack of write activity on this client.
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/grant_shrink_interval
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of seconds with no write activity for this target
|
|
to start releasing dirty grant back to the server.
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/destroys_in_flight
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of DESTROY RPCs currently in flight to this target.
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/lockless_truncate
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls whether lockless truncate RPCs are allowed to this
|
|
target.
|
|
Lockless truncate causes server to perform the locking which
|
|
is beneficial if the truncate is not followed by a write
|
|
immediately.
|
|
1: enable ; 0: disable (default)
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/max_dirty_mb
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls how much dirty data this client can accumulate
|
|
for this target. This is orthogonal to dirty grant and is
|
|
a hard limit even if the server would allow a bigger dirty
|
|
cache.
|
|
While allowing higher dirty cache is beneficial for write
|
|
performance, flushing write cache takes longer and as such
|
|
the node might be more prone to OOMs.
|
|
Having this value set too low might result in not being able
|
|
to sent too many parallel WRITE RPCs.
|
|
Default: 32
|
|
|
|
What: /sys/fs/lustre/osc/{connection_name}/resend_count
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Controls how many times to try and resend RPCs to this target
|
|
that failed with "recoverable" status, such as EAGAIN,
|
|
ENOMEM.
|
|
|
|
What: /sys/fs/lustre/lov/{connection_name}/numobd
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of OSC targets managed by this LOV instance.
|
|
|
|
What: /sys/fs/lustre/lov/{connection_name}/activeobd
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of OSC targets managed by this LOV instance that are
|
|
actually active.
|
|
|
|
What: /sys/fs/lustre/lmv/{connection_name}/numobd
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of MDC targets managed by this LMV instance.
|
|
|
|
What: /sys/fs/lustre/lmv/{connection_name}/activeobd
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Number of MDC targets managed by this LMV instance that are
|
|
actually active.
|
|
|
|
What: /sys/fs/lustre/lmv/{connection_name}/placement
|
|
Date: May 2015
|
|
Contact: "Oleg Drokin" <oleg.drokin@intel.com>
|
|
Description:
|
|
Determines policy of inode placement in case of multiple
|
|
metadata servers:
|
|
CHAR - based on a hash of the file name used at creation time
|
|
(Default)
|
|
NID - based on a hash of creating client network id.
|