@@ -177,11 +177,11 @@ Options for the LizardFS FSAL part of the ganesha.conf file
177177+----------------------------------+-----+---------+-------------+---------------------------------------------------------------+
178178| io_retries | 0 | 1024 | 30 | I/O retries connecting to LizardFS |
179179+----------------------------------+-----+---------+-------------+---------------------------------------------------------------+
180- | chunkserver_round_time_ms | 0 | 65536 | 200 | |
180+ | chunkserver_round_time_ms | 0 | 65536 | 200 | |
181181+----------------------------------+-----+---------+-------------+---------------------------------------------------------------+
182182| chunkserver_connect_timeout_ms | 0 | 65535 | 2000 | Time after which a chunkserver connection is defined dead |
183183+----------------------------------+-----+---------+-------------+---------------------------------------------------------------+
184- | chunkserver_wave_read_timeout_ms | 0 | 65535 | 500 | |
184+ | chunkserver_wave_read_timeout_ms | 0 | 65535 | 500 | Timeout for executing each wave of a read operation |
185185+----------------------------------+-----+---------+-------------+---------------------------------------------------------------+
186186| cache_expiration_time_ms | 0 | 65536 | 1000 | How long till chunks get thrown out of the cache |
187187+----------------------------------+-----+---------+-------------+---------------------------------------------------------------+
@@ -195,7 +195,7 @@ Options for the LizardFS FSAL part of the ganesha.conf file
195195+----------------------------------+-----+---------+-------------+---------------------------------------------------------------+
196196| chunkserver_write_timeout_ms | 0 | 60000 | 5000 | How long to wait for the chunkserver to complete a write cycle|
197197+----------------------------------+-----+---------+-------------+---------------------------------------------------------------+
198- | cache_per_inode_percentage | 0 | 80 | 25 | |
198+ | cache_per_inode_percentage | 0 | 80 | 25 | Max. percentage of write cache per single inode | |
199199+----------------------------------+-----+---------+-------------+---------------------------------------------------------------+
200200| symlink_cache_timeout_s | 0 | 60000 | 3600 | How long to wait for a response from the symlink cache in sec.|
201201+----------------------------------+-----+---------+-------------+---------------------------------------------------------------+
@@ -213,5 +213,32 @@ Options for the LizardFS FSAL part of the ganesha.conf file
213213The **name ** value has to be set to **LizardFS ** or nfs-ganesha will not choose
214214the LizardFS FSAL and will not be able to use LizardFS as a backend.
215215
216+
217+
218+ Providing HA to the NFS Ganesha Plugin
219+ ======================================
220+
221+ For commercial customers HA is provided by the uraft package. Installation and
222+ basic setup should be done according to the instrauctions in
223+ :ref: ’lizardfs_ha_cluster’ .
224+
225+ You should take into account that you need one uraft per service, so if you want
226+ your masters and your ganesha metadata nodes on the same boxes, you need to
227+ install uraft once for the masters and once for the nfs metadata servers. In
228+ such a setup it is vital to make sure that the two instances listen on
229+ different ports and manage different ip addresses.
230+
231+ Assuming that your master server uraft seetup is setup according to the defaults
232+ in the :ref: ’lizardfs_ha_cluster’ chapter, we would suggest to use port 9527 and
233+ 9528 as port and status port in your uraft setup for NFS as follows:
234+
235+ URAFT_PORT = 9527
236+ URAFT_STATUS_PORT: 9528
237+
238+ and make sure that you use a designated floating IP for your NFS services.
239+
240+ All other settings should be as described in :ref: ’lizardfs_ha_cluster’ .
241+
242+
243+
216244.. _TODO : add descriptions for undescribed FSAL options
217- .. _TODO : add description for uraft-nfs
0 commit comments