Just like eglynn suggested in his comment of https://review.openstack.org/#/c/43073/, as I quote here:
However, I would like to revisit the issue of fine-grained versus bulk SNMP commands at some point in the future. IIUC the interaction could be made significantly less chatty if the bulk approach was used. I take your point that this is not supported by older SNMP versions, but I wonder could we always first assume that the bulk mode is supported by each host, and only revert to fine-grained commands if the bulk version fails in a way that suggests that mode is not supported by the SNMP daemon running on that host (and cache that per-host status).
Please note that in older version of SNMP(version<2c), bulkCmd is not supported. So we need to take that into account too.
Reviewed: https:/ /review. openstack. org/108938 /git.openstack. org/cgit/ openstack/ ceilometer/ commit/ ?id=e7046c44c01 596abae9ef867cf 0cd95a51919dac
Committed: https:/
Submitter: Jenkins
Branch: master
commit e7046c44c01596a bae9ef867cf0cd9 5a51919dac
Author: Lianhao Lu <email address hidden>
Date: Fri Jul 25 13:19:32 2014 +0800
Added new hardware inspector interface
Added a new hardware inspector interface: inspect_generic and its snmp
implementation. The old interface will be removed later.
Partial implements: blueprint snmp-improvement.
Partial-Bug: 1286132
Change-Id: I847cfdad5b6571 c4d066250617050 a8f0b1cfbc4