d4ff29f6c5
hyper-v fix when management interface is not part of virtual switch
...
2
formatting fix
2019-08-22 07:54:50 +02:00
cd3b39ad17
commit generated code
2019-08-12 15:55:43 -07:00
a6830f8348
Move 'ui' variable closer to invocation.
2019-07-31 14:08:40 -04:00
e362d2f3ee
Use 'log.Printf()' for StepDownload debug instead of 'ui.Say()'.
2019-07-31 14:02:57 -04:00
90c5da40f2
implement abort logic and printing for the askstep implementation as well
2019-07-26 12:24:03 -07:00
54034689ef
On abort, return gracefully rather than exiting so that the subprocess doesn't unexpectedly disconnect from the parent and cause a confusing EOF error in the logs
2019-07-25 16:32:16 -07:00
216afbaa8b
Fixed various typos
...
Just a bunch of nit-picky typo fixes.
2019-07-25 06:49:51 +02:00
e9549d2688
add test for local fs download
2019-07-22 09:17:48 -07:00
197a283b5e
actually check for filepath being present
2019-07-19 10:35:50 -07:00
64f16474f8
fix double slash conundrum
2019-07-19 10:16:40 -07:00
aef77771c0
Correct HTTPIP == FALSE on HyperV builder
2019-07-17 03:41:41 -05:00
d6d4eb2087
fix some tests and some config behavior to prevent null dereference errors and incorrect precedence between iso checksum and iso checksum url
2019-06-28 15:29:39 -07:00
998ef4053f
parse checksum file so that we can use the checksums inside in the vmware-esxi builder stepuploadiso
2019-06-28 14:35:43 -07:00
88d589cc88
enable smb share double backslash pathing too
2019-06-26 13:43:05 -07:00
12471a8b06
use file getter for smb shares
2019-06-26 13:21:49 -07:00
8d25cc578f
Merge pull request #7746 from hashicorp/fix_7671
...
Abort hyperv build if there's a name collision
2019-06-14 10:58:22 -07:00
2e821da84b
check to make sure a vm-name isn't already in use before trying to launch a vm with said name.
2019-06-13 14:09:45 -07:00
ca33f8bc5c
Revert "Backoff.Linear: panic when InitialBackoff > MaxBackoff"
...
This reverts commit 39cfacd5fa
2019-06-11 12:53:06 +02:00
39cfacd5fa
Backoff.Linear: panic when InitialBackoff > MaxBackoff
...
this probably means there's a configuration issue.
Since this struct is mainly set manually from code, I think it is okay to panic here.
2019-06-11 12:41:21 +02:00
36e4eaff99
document retry.Backoff better
2019-06-11 12:20:00 +02:00
8519da6ce9
common/net: Cleanup cache of used port after closing
2019-05-06 22:11:52 +01:00
b329073e54
remove trying port log as retry code will show errors
...
less logs !
2019-05-01 13:04:51 +02:00
38f789eedc
add error types to test for
2019-05-01 13:01:25 +02:00
14f2d1c132
ListenRangeConfig:Listen: use the retry pkg to wait a bit in between opens
2019-05-01 12:27:32 +02:00
e14d1b8b0b
add a timeout trying to open a random port
2019-05-01 12:01:34 +02:00
2fd7adffc4
add tests for ListenRangeConfig.Listen
2019-05-01 11:46:48 +02:00
d9e5145de9
Shadow the flock pkg & add a noop filelock so that at least the solaris binary can be built
...
Without this the following error occurs:
$ GOOS=solaris go build .
# github.com/hashicorp/packer/vendor/github.com/gofrs/flock
vendor/github.com/gofrs/flock/flock_unix.go:28:22: undefined: syscall.LOCK_EX
vendor/github.com/gofrs/flock/flock_unix.go:39:22: undefined: syscall.LOCK_SH
vendor/github.com/gofrs/flock/flock_unix.go:56:12: undefined: syscall.Flock
vendor/github.com/gofrs/flock/flock_unix.go:66:12: undefined: syscall.Flock
vendor/github.com/gofrs/flock/flock_unix.go:96:12: undefined: syscall.Flock
vendor/github.com/gofrs/flock/flock_unix.go:96:42: undefined: syscall.LOCK_UN
vendor/github.com/gofrs/flock/flock_unix.go:118:21: undefined: syscall.LOCK_EX
vendor/github.com/gofrs/flock/flock_unix.go:130:21: undefined: syscall.LOCK_SH
vendor/github.com/gofrs/flock/flock_unix.go:149:9: undefined: syscall.Flock
vendor/github.com/gofrs/flock/flock_unix.go:149:44: undefined: syscall.LOCK_NB
vendor/github.com/gofrs/flock/flock_unix.go:149:44: too many errors
2019-04-30 16:01:23 +02:00
5d9d43c01d
add tests for ListenRangeConfig.Listen
2019-04-30 15:42:59 +02:00
fd63ec9a6c
ListenRangeConfig.Listen: move trying port log after fslock trylock for less logs
2019-04-30 15:35:22 +02:00
79ae62801f
ListenRangeConfig: default network ( protocol ) to tcp
2019-04-30 14:55:38 +02:00
65be2be38b
step_download: force copy of local files only on windows
2019-04-29 09:39:21 +02:00
200e26ea85
Step download: always copy local files
...
instead of symlinking to fix #7534 . The longer term fix for this would be to change the go-getter so that it can leave the source file where it is & tell us where it is.
We will do this when the right time comes.
2019-04-26 10:35:18 +02:00
9f1136db77
retry: encapsulate & return the last seen error in a RetryExhaustedError
2019-04-24 14:19:23 +02:00
e69d95eb37
when retries were exhausted in the retry Run, we were returning nil instead of an error.
2019-04-23 13:48:30 -07:00
5f0259fe89
step download: don't fail when checksum_type is not set
2019-04-16 18:09:54 +02:00
4dc12701d3
step_download: display clearer error messages, also when the checksum is not set
...
Usually the builder validation should error when the checksum is empty, but for the virtualbox-ovf builder this is not validated. Which is something I didn't see when refactoring for go-getter incorporation/refactor.
2019-04-15 14:34:09 +02:00
78cf1405ab
Merge pull request #7480 from BobSilent/fix-hyperv-multiple-hostadapter-same-switch
...
Fixes "Host IP for the HyperV machine: False" if more than one VMNetworkAdapter is found
2019-04-12 14:06:46 -07:00
9ac8d8aed8
step download: ovf files usually point to a file in the same directory, using them in place is the only way
2019-04-12 16:51:57 +02:00
63f5bbbc73
use our own copy of iochan.LineReader
2019-04-11 14:25:24 +02:00
d72040f4fa
move retry code into the common/retry pkg and make retry context aware
2019-04-09 17:46:38 +02:00
5b289f67fe
Update hyperv.go
...
maybe also a fix for issue #5023
at least I got same error like there, incl. VM response
==> hyperv-iso: Host IP for the HyperV machine: False
although #4947 is marked as duplicate but shows different error output
==> hyperv-iso: Error getting host adapter ip address: PowerShell error: Get-VMNetworkAdapter : No network adapter is found with the given input.
2019-04-09 07:36:52 +02:00
f555e7a9f2
allow a provisioner to timeout
...
* I had to contextualise Communicator.Start and RemoteCmd.StartWithUi
NOTE: Communicator.Start starts a RemoteCmd but RemoteCmd.StartWithUi will run the cmd and wait for a return, so I renamed StartWithUi to RunWithUi so that the intent is clearer.
Ideally in the future RunWithUi will be named back to StartWithUi and the exit status or wait funcs of the command will allow to wait for a return. If you do so please read carrefully https://golang.org/pkg/os/exec/#Cmd.Stdout to avoid a deadlock
* cmd.ExitStatus to cmd.ExitStatus() is now blocking to avoid race conditions
* also had to simplify StartWithUi
2019-04-08 20:09:21 +02:00
a81abd297b
Merge remote-tracking branch 'origin/master' into context_provisioner
2019-04-08 20:09:01 +02:00
3e72e65a6c
Merge pull request #7456 from hashicorp/do_5770
...
introduce the clean_resource_name to clean image/var names
2019-04-05 09:12:20 -07:00
217a93fc30
fix build_test.go on windows. This code needs to not be run in parallel, or else there can be race conditions with the post-processors all trying to access the same file at once. Additionally, I changed the test template itself so that it is valid code on either bash or cmd. Finally, I found and fixed a small bug in naming the file extensions in shell local inline scripts.
2019-04-04 14:57:56 -07:00
c4f3dccc14
rename interpolation context from ctx to ictx and contexts to ctx to avoid conflicts
2019-04-03 15:56:15 +02:00
82c4b76639
[Builder|Build|PostProcessor|Provisioner|Hook]Server: context cancel using a RPC cancel method
2019-04-03 15:55:55 +02:00
14048b1e11
Revert "WIP: go towards context cancelled rpc pkg"
...
This reverts commit e4609b4a6a4ba2ec76a45ad67d50d180f1a914f9.
2019-04-03 15:55:55 +02:00
9e3a179c0d
WIP: go towards context cancelled rpc pkg
2019-04-03 15:55:55 +02:00
829851fc8a
change hooks to be passed a context for cancellation
...
we have to as it is what calls our provisioners
2019-04-03 15:55:54 +02:00
Marcin Bojko
Megan Marsh
Stephen Fox
Christian Muehlhaeuser
WilliamB
Adrien Delorme
Radek Simko
BobSilent