Optimize the speed of creating disks #100
Conversation
| # process each storage object added to the gateway, and map to the tpg | ||
| for stg_object in lio_root.storage_objects: | ||
|
|
||
| for tpg in self.tpg_list: | ||
| self.logger.debug("processing tpg{}".format(tpg.tag)) | ||
|
|
||
| if not self.lun_mapped(tpg, stg_object): |
There was a problem hiding this comment.
With this patch what is the creation time now?
What part of lun_mapped was the issue? Was it constantly scanning the lun dir in configfs or something else?
| # process each storage object added to the gateway, and map to the tpg | ||
| for stg_object in lio_root.storage_objects: | ||
|
|
||
| for tpg in self.tpg_list: | ||
| self.logger.debug("processing tpg{}".format(tpg.tag)) | ||
|
|
||
| if not self.lun_mapped(tpg, stg_object): | ||
| if not tpg_stg_object_dict.has_key(str(tpg.tag) + "-" + stg_object.name): |
There was a problem hiding this comment.
I don't think we actually need to call lun_mapped for this as LUN() checks if the object is already mapped for us. We could just do:
for stg_object in lio_root.storage_objects:
for tpg in self.tpg_list:
try:
mapped_lun = LUN(tpg, lun=lun_id, storage_object=stg_object)
....
except RTSLibError as err:
if "already exists in configFS" not in str(err):
self.logger.error("LUN mapping failed: {}".format(err))
self.error = True
self.error_msg = str(err)
return
#Already created. Ignore and loop to the next tpg.
continue
.....
Oh yeah, could your post your patch to this GH:
https://github.com/ceph/ceph-iscsi
instead of ceph-iscsi-config/cli?
ceph-iscsi is the repo that will be used going forward. ceph-iscsi-config/cli is more only used for a older version of the tools/lib.
There was a problem hiding this comment.
when catch a exception, configuration will be cleaned and the program exit.
|
Closing this PR. We are using: |
#99