The problem is: There is no mechanism to detect block device setup failure
Network devices have the same problem, and are fixed with this too.
I handling this problem in the way suggested by aliguori:
- hotplug scripts write a "hotplug-status" node on store
- Xend DevController.createDevice() check verify this node and return success
or throw an exception on failure.
- If no changes in "hotplug-status" node after DEVICE_CREATE_TIMEOUT seconds
Xend throw an exception showing the problem with hotplug scripts.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
list is passed rather than a single string. This expands the config-file
interface so that parameters may be passed to the network scripts from the
config file.
Fix XendRoot._logError (missing asterisk before args parameter).
In XendRoot, allow an empty configuration file. This makes it easier to unit
test code reliant on XendRoot.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
for the command line parsing code in xm create. To make it easier to test,
create.py has changed in structure around the main(argv) function, adding a new
parseCommandLine function.
Remove a large number of useless parameters, where the opts structure was being
passed all over the place, just to eventually log through it. Instead two
methods -- err and warn -- have been broken out from the Opts class, and there
is now no need to pass the opts value everywhere.
Change the parsing inside make_config to remove lots of cut-and-pasted cruft.
Change the documentation to the vif command to indicate that the bridge can be
auto-detected.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
This was passing down to dhclient which was keeping the 8001 port
open and preventing xend restarting.
Also fixed a case typo in setCloExec
Signed-off-by: James Bulpin <james@xensource.com>
be refused, with the domain being destroyed instead.
Further fix for reboot logic when using on_reboot=rename-restart: the incorrect
node was being removed once restart has completed, leaving Xend to conclude
later that it had crashed in the middle of a restart.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
Reject the removal of -Werror from the xentop Makefile.
Reject the #include of xs.h inside xentop -- there is no need for it.
Have XendDomainInfo write domain names into /local/domain (which used to happen,
but hasn't for a little while).
Remove the dead code from xentop.c.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
It doesn't make sense to undo the network configuration when stopping
xend since that would cut off all VMs from the network. Undoing the
network setup should be an option independent of xend stop.
Signed-off-by: Christian Limpach <Christian.Limpach@cl.cam.ac.uk>
whether a domain has been introduced to the store. Added a separate message
XS_IS_DOMAIN_INTRODUCED and API for that (xs_is_domain_introduced) to determine
whether the domain has really been introduced. This change means that the
tools can determine the correct domain path earlier in the domain creation
process, which is particularly a factor with live migration, as it allows us
to create the devices earlier in the process, and unpause the new domain before
performing the introduce. Until recently we already had these features, but
the simplification of the interface between xend and xenstored caused breakage.
No longer clear out the domain path when a domain is introduced -- this was a
hack to work around the recent problematic semantics of GetDomainPath.
Do not write the contents of the info block to the store. All the configuration
info is written to the /vm path, and anything else in the info block is either
dealt with explicitly or is ephemeral and has no place in the store.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
/local/domain/0/backend/<devclass>/<dom>/<devid>/error, write them instead to
/local/domain/0/error/backend/<yada>. This is not the best place for them
perhaps, but it moves them out of the backend directory, on which the drivers
have a watch. This fixes the problem whereby writing an error will trigger a
watch, causing the error message to be written again, and repeat. Fixes bug
#286.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
for some arguemnt like 'localtime', 'isa', device model need an argument
"-localtime", instead of "-localtime 1"
Signed-off-by: Xiaofeng Ling <xiaofeng.ling@intel.com>
1) enable ACPI in VMXAssist by default.
2) do nothing when emulating wbinvd instruction in VMXAssist.
3) use MACRO in cpu number setting/getting code.
4) remove useless parameter flags from xc_vmx_build.
5) remove usage of dsi.v_start in xc_vmx_build when not handling
vmxloader elf.
6) unmap shared_page_va when destroy VMX domain.
7) change virtual_platform_def to vmx_platform.
8) remove useless code in vmx_setup_platform.
9) change parameter of vmx_setup_platform from vcpu to domain.
10) in xen HV, vmx domain get domain processor number from the reserved
E820 area.
11) in domain.c, don't define some functions when on i386 platform or no
CONFIG_VMX.
Also removed some ugly tailing space.
Signed-off-by: Xin Li <xin.b.li@intel.com>
disabled by workaround 7468:17a9f111fa93. We additionally need to set the
FD_CLOEXEC flag on the status fd given to SrvServer when spawning the network
script, as at least on some platforms this causes xend to fail to start
properly.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
hotplugging stuff was introduced (block-*, console-port-base, console-address)
and introducing entries for options that have been present for ages
(xend-{http,unix,relocation}-server, xend-unix-path, xend-relocation-address,
enable-dump). Remove vif-antispoof, as Vifctl no longer passes this option
down.
Made the xend-relocation-server default to no, as this is a potential security
hole and so must be turned on explicitly.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
antispoof parameters from Vifctl.network -- these are already handled by the
script, so it is redundant to have a separate override for these parameters
coming from outside. Don't use xen.util.process.runscript, because we don't
need to return the output, and therefore can just use os.spawnl.
Remove XendRoot.get_vif_antispoof, matching change above.
Prefix XendRoot.network_script_dir onto the return value for
XendRoot.get_network_script(), removing that burden from callers.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
Store the uuid in string form inside XendDomainInfo, to aid this.
Add restart-squelching logic to ensure that runaway domains to restart
indefinitely. The logic here could be cleverer, say including a backoff for
retrying, in case a network block device has gone away, but for now this will
do to stop the pathological behaviour. Fixes bug #275.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
This is Kurt Garloff's reworked network-bridge script:
* we got rid of ifconfig
* it works for netdev != eth0
* arp on and off are symmetric as are ifdown and ifup
* ifup will be passed the ifcfg config file name if needed
(the ifup may otherwise figure that the veth0 hardware is
NOT the same as the original ${netdev} and not use the same
config -- this happens on SUSE. Charles Coffing tracked this
one down.)
Plus Kurt's avoid-dash patch:
the network setup scripts on SUSE have trouble with the bridge
name xen-br0; they don't expect the '-'.
Arguably this should be fixed.
But I assume there's more scripts out there which may not like it,
so I suggest the following patch to rename xen-br0 to xenbr0.
Plus Charles Duffy's patch to support multiple bridges:
The attached patch allows the network-bridge script to be used to
generate multiple bridges corresponding to different physical
interfaces. It adds a new parameter, "vifnum", used to refer both to
the loopback interface to be used and to set defaults regarding the
physical interface and bridge name.
Thus, if one wishes to start xenbr0 on eth0 and xenbr1 on eth1, one
need only call:
network-bridge start ## vifnum is 0 by default
network-bridge start vifnum=1
...well, that and set loopback.nloopbacks=2 in the Dom0 kernel
parameters.
Plus renaming of virtnum to vifnum in Charles' patch, as requested by Ian Pratt.
Plus a fix to DevController to allocate vif IDs starting from 0 (i.e. vif2.0
is now domain 2's first vif, as opposed to vif2.1 in the recent past).
Plus tidying up inside network-bridge using some helper variables.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
recreation semantics.
Remove the unpause at the end of xc_linux_restore, and move it to
XendDomainInfo. This is necessary because xenstored now allocates the domain
path when the domain is introduced, which means that the new domain cannot
start running until that introduce is performed and the new devices
configured.
Give restore a separate completion phase in which domain details are stored.
This is required because the domain path is no longer available until after
the introduceDomain call.
TODO: Split the domain introduction into two so that the domain path is
available earlier. At the moment, the domain <-> store channel details are
passed in to xenstored when the domain is introduced, but in the case of
restore it is necessary to wait until the restore is completed before the
channel MFN is available.
Change the interface between XendDomainInfo and XendCheckpoint/image to not
have hideous callbacks through setConsoleRef and setStoreRef. Instead,
image.createImage explicitly returns those values, and XendCheckpoint passes
them through to completeRestore.
Move the purging of the domain path corresponding to a new domain from Xend
to xenstored, since xenstored is now in charge of this path. With the domain
path creation moved to xenstored, Xend cannot remove the path, because watches
may have fired on it already.
Fix the printf statement in xenstored in verbose mode that details the
messages being written. This statement was printing the buffer using %s, but
this buffer has an explicit length field, so we were seeing garbage after the
correct details.
Signed-off-by: Ewan Mellor <ewan@xensource.com>
simply creates a CPU on every physical CPU. Other domains
read the initial VCPU configuration from xenstore.
Signed-off-by: Keir Fraser <keir@xensource.com>
