From 070e998b40b5764a5f348b4e692ed9c00207799f Mon Sep 17 00:00:00 2001
From: Iustin Pop <iustin@google.com>
Date: Wed, 10 Jun 2009 10:46:08 +0200
Subject: [PATCH] Convert node_info rpc to new style result
This patch also does some cleanup and enforces valid results (with
proper type, i.e. int for memory/disk values) from remote node,
otherwise we handle the result as failure.
We do this so that we can remove custom processing in rpc.py which is
very old and doesn't play with the new style result (and has bit me by
modifying the result received from the remote node).
After this change, the node_info rpc becomes a normal one.
Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>
---
lib/backend.py | 2 +-
lib/cmdlib.py | 76 ++++++++++++++++++++++++++++++--------------------
lib/rpc.py | 21 ++------------
3 files changed, 48 insertions(+), 51 deletions(-)
diff --git a/lib/backend.py b/lib/backend.py
index a34364936..f087c394b 100644
--- a/lib/backend.py
+++ b/lib/backend.py
@@ -367,7 +367,7 @@ def GetNodeInfo(vgname, hypervisor_type):
finally:
f.close()
- return outputarray
+ return True, outputarray
def VerifyNode(what, cluster_name):
diff --git a/lib/cmdlib.py b/lib/cmdlib.py
index 2d51ad255..5e9774915 100644
--- a/lib/cmdlib.py
+++ b/lib/cmdlib.py
@@ -1999,8 +1999,8 @@ class LUQueryNodes(NoHooksLU):
self.cfg.GetHypervisorType())
for name in nodenames:
nodeinfo = node_data[name]
- if not nodeinfo.failed and nodeinfo.data:
- nodeinfo = nodeinfo.data
+ if not nodeinfo.RemoteFailMsg() and nodeinfo.payload:
+ nodeinfo = nodeinfo.payload
fn = utils.TryConvert
live_data[name] = {
"mtotal": fn(int, nodeinfo.get('memory_total', None)),
@@ -2833,15 +2833,17 @@ def _CheckNodeFreeMemory(lu, node, reason, requested, hypervisor_name):
"""
nodeinfo = lu.rpc.call_node_info([node], lu.cfg.GetVGName(), hypervisor_name)
- nodeinfo[node].Raise()
- free_mem = nodeinfo[node].data.get('memory_free')
+ msg = nodeinfo[node].RemoteFailMsg()
+ if msg:
+ raise errors.OpPrereqError("Can't get data from node %s: %s" % (node, msg))
+ free_mem = nodeinfo[node].payload.get('memory_free', None)
if not isinstance(free_mem, int):
raise errors.OpPrereqError("Can't compute free memory on node %s, result"
- " was '%s'" % (node, free_mem))
+ " was '%s'" % (node, free_mem))
if requested > free_mem:
raise errors.OpPrereqError("Not enough memory on node %s for %s:"
- " needed %s MiB, available %s MiB" %
- (node, reason, requested, free_mem))
+ " needed %s MiB, available %s MiB" %
+ (node, reason, requested, free_mem))
class LUStartupInstance(LogicalUnit):
@@ -4813,19 +4815,19 @@ class LUCreateInstance(LogicalUnit):
self.op.hypervisor)
for node in nodenames:
info = nodeinfo[node]
- info.Raise()
- info = info.data
- if not info:
+ msg = info.RemoteFailMsg()
+ if msg:
raise errors.OpPrereqError("Cannot get current information"
- " from node '%s'" % node)
+ " from node %s: %s" % (node, msg))
+ info = info.payload
vg_free = info.get('vg_free', None)
if not isinstance(vg_free, int):
raise errors.OpPrereqError("Can't compute free disk space on"
" node %s" % node)
- if req_size > info['vg_free']:
+ if req_size > vg_free:
raise errors.OpPrereqError("Not enough disk space on target node %s."
" %d MB available, %d MB required" %
- (node, info['vg_free'], req_size))
+ (node, vg_free, req_size))
_CheckHVParams(self, nodenames, self.op.hypervisor, self.op.hvparams)
@@ -5673,10 +5675,11 @@ class LUGrowDisk(LogicalUnit):
instance.hypervisor)
for node in nodenames:
info = nodeinfo[node]
- if info.failed or not info.data:
+ msg = info.RemoteFailMsg()
+ if msg:
raise errors.OpPrereqError("Cannot get current information"
- " from node '%s'" % node)
- vg_free = info.data.get('vg_free', None)
+ " from node %s:" % (node, msg))
+ vg_free = info.payload.get('vg_free', None)
if not isinstance(vg_free, int):
raise errors.OpPrereqError("Can't compute free disk space on"
" node %s" % node)
@@ -6115,9 +6118,15 @@ class LUSetInstanceParams(LogicalUnit):
instance.hypervisor)
nodeinfo = self.rpc.call_node_info(mem_check_list, self.cfg.GetVGName(),
instance.hypervisor)
- if nodeinfo[pnode].failed or not isinstance(nodeinfo[pnode].data, dict):
+ pninfo = nodeinfo[pnode]
+ msg = pninfo.RemoteFailMsg()
+ if msg:
# Assume the primary node is unreachable and go ahead
- self.warn.append("Can't get info from primary node %s" % pnode)
+ self.warn.append("Can't get info from primary node %s: %s" %
+ (pnode, msg))
+ elif not isinstance(pninfo.payload.get('memory_free', None), int):
+ self.warn.append("Node data from primary node %s doesn't contain"
+ " free memory information" % pnode)
elif instance_info.RemoteFailMsg():
self.warn.append("Can't get instance runtime information: %s" %
instance_info.RemoteFailMsg())
@@ -6130,19 +6139,24 @@ class LUSetInstanceParams(LogicalUnit):
# and we have no other way to check)
current_mem = 0
miss_mem = (be_new[constants.BE_MEMORY] - current_mem -
- nodeinfo[pnode].data['memory_free'])
+ pninfo.payload['memory_free'])
if miss_mem > 0:
raise errors.OpPrereqError("This change will prevent the instance"
" from starting, due to %d MB of memory"
" missing on its primary node" % miss_mem)
if be_new[constants.BE_AUTO_BALANCE]:
- for node, nres in nodeinfo.iteritems():
+ for node, nres in nodeinfo.items():
if node not in instance.secondary_nodes:
continue
- if nres.failed or not isinstance(nres.data, dict):
- self.warn.append("Can't get info from secondary node %s" % node)
- elif be_new[constants.BE_MEMORY] > nres.data['memory_free']:
+ msg = nres.RemoteFailMsg()
+ if msg:
+ self.warn.append("Can't get info from secondary node %s: %s" %
+ (node, msg))
+ elif not isinstance(nres.payload.get('memory_free', None), int):
+ self.warn.append("Secondary node %s didn't return free"
+ " memory information" % node)
+ elif be_new[constants.BE_MEMORY] > nres.payload['memory_free']:
self.warn.append("Not enough memory to failover instance to"
" secondary node %s" % node)
@@ -6920,24 +6934,24 @@ class IAllocator(object):
}
if not ninfo.offline:
- nresult.Raise()
- if not isinstance(nresult.data, dict):
- raise errors.OpExecError("Can't get data for node %s" % nname)
+ msg = nresult.RemoteFailMsg()
+ if msg:
+ raise errors.OpExecError("Can't get data for node %s: %s" %
+ (nname, msg))
msg = node_iinfo[nname].RemoteFailMsg()
if msg:
raise errors.OpExecError("Can't get node instance info"
" from node %s: %s" % (nname, msg))
- remote_info = nresult.data
+ remote_info = nresult.payload
for attr in ['memory_total', 'memory_free', 'memory_dom0',
'vg_size', 'vg_free', 'cpu_total']:
if attr not in remote_info:
raise errors.OpExecError("Node '%s' didn't return attribute"
" '%s'" % (nname, attr))
- try:
- remote_info[attr] = int(remote_info[attr])
- except ValueError, err:
+ if not isinstance(remote_info[attr], int):
raise errors.OpExecError("Node '%s' returned invalid value"
- " for '%s': %s" % (nname, attr, err))
+ " for '%s': %s" %
+ (nname, attr, remote_info[attr]))
# compute memory used by primary instances
i_p_mem = i_p_up_mem = 0
for iinfo, beinfo in i_list:
diff --git a/lib/rpc.py b/lib/rpc.py
index 7d6b26614..2e5d638ec 100644
--- a/lib/rpc.py
+++ b/lib/rpc.py
@@ -647,25 +647,8 @@ class RpcRunner(object):
memory information
"""
- retux = self._MultiNodeCall(node_list, "node_info",
- [vg_name, hypervisor_type])
-
- for result in retux.itervalues():
- if result.failed or not isinstance(result.data, dict):
- result.data = {}
- if result.offline:
- log_name = None
- else:
- log_name = "call_node_info"
-
- utils.CheckDict(result.data, {
- 'memory_total' : '-',
- 'memory_dom0' : '-',
- 'memory_free' : '-',
- 'vg_size' : 'node_unreachable',
- 'vg_free' : '-',
- }, log_name)
- return retux
+ return self._MultiNodeCall(node_list, "node_info",
+ [vg_name, hypervisor_type])
def call_node_add(self, node, dsa, dsapub, rsa, rsapub, ssh, sshpub):
"""Add a node to the cluster.
--
GitLab