vdsm-patches May 2015

vdsm-patches@lists.fedorahosted.org

55 participants
833 discussions

Change in vdsm[master]: spbackends: do not set spmRole on forceFreeSpm
by Federico Simoncelli 19 Oct '15

19 Oct '15

Federico Simoncelli has uploaded a new change for review. Change subject: spbackends: do not set spmRole on forceFreeSpm ...................................................................... spbackends: do not set spmRole on forceFreeSpm Setting the spmRole to SPM_FREE on forceFreeSpm is harmful. In fact the release of the spm role (SPM_ACQUIRED) should go through the stopSpm procedure where the master filesystem is unmounted, etc. Change-Id: I2bd9a0d9749e49a97a31c535c92dd242eb8f74ec Signed-off-by: Federico Simoncelli <fsimonce(a)redhat.com> --- M vdsm/storage/spbackends.py 1 file changed, 0 insertions(+), 1 deletion(-) git pull ssh://gerrit.ovirt.org:29418/vdsm refs/changes/18/27318/1 diff --git a/vdsm/storage/spbackends.py b/vdsm/storage/spbackends.py index 86714a3..dddb749 100644 --- a/vdsm/storage/spbackends.py +++ b/vdsm/storage/spbackends.py @@ -327,7 +327,6 @@ # DO NOT USE, STUPID, HERE ONLY FOR BC # TODO: SCSI Fence the 'lastOwner' self.setSpmStatus(LVER_INVALID, SPM_ID_FREE, __securityOverride=True) - self.pool.spmRole = SPM_FREE @classmethod def _getPoolMD(cls, domain): -- To view, visit http://gerrit.ovirt.org/27318 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: newchange Gerrit-Change-Id: I2bd9a0d9749e49a97a31c535c92dd242eb8f74ec Gerrit-PatchSet: 1 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: Federico Simoncelli <fsimonce(a)redhat.com>

6 11

Change in vdsm[master]: [wip] sdcache: avoid extra refresh due samplingmethod
by Federico Simoncelli 19 Oct '15

19 Oct '15

Federico Simoncelli has uploaded a new change for review. Change subject: [wip] sdcache: avoid extra refresh due samplingmethod ...................................................................... [wip] sdcache: avoid extra refresh due samplingmethod In order to avoid an extra iscsi rescan (symptomatic of samplingmethod) an additional lock has been introduced to queue the requests when the storage is flagged as stale. Bug-Url: https://bugzilla.redhat.com/show_bug.cgi?id=870768 Change-Id: If178a8eaeb94f1dfe9e0957036dde88f6a22829c Signed-off-by: Federico Simoncelli <fsimonce(a)redhat.com> --- M vdsm/storage/sdc.py 1 file changed, 25 insertions(+), 26 deletions(-) git pull ssh://gerrit.ovirt.org:29418/vdsm refs/changes/74/9274/1 diff --git a/vdsm/storage/sdc.py b/vdsm/storage/sdc.py index f2f4534..978e3fa 100644 --- a/vdsm/storage/sdc.py +++ b/vdsm/storage/sdc.py @@ -62,32 +62,27 @@ STORAGE_UPDATED = 0 STORAGE_STALE = 1 - STORAGE_REFRESHING = 2 def __init__(self, storage_repo): - self._syncroot = threading.Condition() + self._syncDomain = threading.Condition() + self._syncRefresh = threading.Lock() self.__domainCache = {} self.__inProgress = set() self.__staleStatus = self.STORAGE_STALE self.storage_repo = storage_repo def invalidateStorage(self): - with self._syncroot: - self.__staleStatus = self.STORAGE_STALE + self.log.debug("The storages have been invalidated") + self.__staleStatus = self.STORAGE_STALE @misc.samplingmethod def refreshStorage(self): - self.__staleStatus = self.STORAGE_REFRESHING - + # We need to set the __staleStatus value at the beginning because we + # want to keep track of the future invalidateStorage calls that might + # arrive during the rescan procedure. + self.__staleStatus = self.STORAGE_UPDATED multipath.rescan() lvm.invalidateCache() - - # If a new invalidateStorage request came in after the refresh - # started then we cannot flag the storages as updated (force a - # new rescan later). - with self._syncroot: - if self.__staleStatus == self.STORAGE_REFRESHING: - self.__staleStatus = self.STORAGE_UPDATED def produce(self, sdUUID): domain = DomainProxy(self, sdUUID) @@ -98,7 +93,7 @@ return domain def _realProduce(self, sdUUID): - with self._syncroot: + with self._syncDomain: while True: domain = self.__domainCache.get(sdUUID) @@ -109,25 +104,29 @@ self.__inProgress.add(sdUUID) break - self._syncroot.wait() + self._syncDomain.wait() try: - # If multiple calls reach this point and the storage is not - # updated the refreshStorage() sampling method is called - # serializing (and eventually grouping) the requests. - if self.__staleStatus != self.STORAGE_UPDATED: - self.refreshStorage() + # Here we cannot take full advantage of the refreshStorage + # samplingmethod since we might be scheduling an unneeded + # extra rescan. We need an additional lock (_syncRefresh) + # to make sure that __staleStatus is taken in account + # (without affecting all the other external refreshStorage + # calls as it would be if we move this check there). + with self._syncRefresh: + if self.__staleStatus != self.STORAGE_UPDATED: + self.refreshStorage() domain = self._findDomain(sdUUID) - with self._syncroot: + with self._syncDomain: self.__domainCache[sdUUID] = domain return domain finally: - with self._syncroot: + with self._syncDomain: self.__inProgress.remove(sdUUID) - self._syncroot.notifyAll() + self._syncDomain.notifyAll() def _findDomain(self, sdUUID): import blockSD @@ -162,16 +161,16 @@ return uuids def refresh(self): - with self._syncroot: + with self._syncDomain: lvm.invalidateCache() self.__domainCache.clear() def manuallyAddDomain(self, domain): - with self._syncroot: + with self._syncDomain: self.__domainCache[domain.sdUUID] = domain def manuallyRemoveDomain(self, sdUUID): - with self._syncroot: + with self._syncDomain: try: del self.__domainCache[sdUUID] except KeyError: -- To view, visit http://gerrit.ovirt.org/9274 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: newchange Gerrit-Change-Id: If178a8eaeb94f1dfe9e0957036dde88f6a22829c Gerrit-PatchSet: 1 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: Federico Simoncelli <fsimonce(a)redhat.com>

4 10

Change in vdsm[master]: sp: load dumped tasks when recovering
by Federico Simoncelli 19 Oct '15

19 Oct '15

Federico Simoncelli has uploaded a new change for review. Change subject: sp: load dumped tasks when recovering ...................................................................... sp: load dumped tasks when recovering Change-Id: I1cd2ea34c2013870b213d8baa471248adabfbbe3 Signed-off-by: Federico Simoncelli <fsimonce(a)redhat.com> --- M vdsm/storage/sp.py 1 file changed, 1 insertion(+), 2 deletions(-) git pull ssh://gerrit.ovirt.org:29418/vdsm refs/changes/02/26902/1 diff --git a/vdsm/storage/sp.py b/vdsm/storage/sp.py index 338232f..e6a6bd0 100644 --- a/vdsm/storage/sp.py +++ b/vdsm/storage/sp.py @@ -288,8 +288,6 @@ self.log.error("Backup domain validation failed", exc_info=True) - self.taskMng.loadDumpedTasks(self.tasksDir) - self.spmRole = SPM_ACQUIRED # Once setSecure completes we are running as SPM @@ -322,6 +320,7 @@ # Restore tasks is last because tasks are spm ops (spm has to # be started) + self.taskMng.loadDumpedTasks(self.tasksDir) self.taskMng.recoverDumpedTasks() self.log.debug("ended.") -- To view, visit http://gerrit.ovirt.org/26902 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: newchange Gerrit-Change-Id: I1cd2ea34c2013870b213d8baa471248adabfbbe3 Gerrit-PatchSet: 1 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: Federico Simoncelli <fsimonce(a)redhat.com>

3 3

Change in vdsm[master]: volume: prepare only one volume on clone
by Federico Simoncelli 19 Oct '15

19 Oct '15

Federico Simoncelli has uploaded a new change for review. Change subject: volume: prepare only one volume on clone ...................................................................... volume: prepare only one volume on clone When we are cloning a volume we need only Change-Id: Idc009fac4dc1a258537b0ffb15bd627680d79330 Signed-off-by: Federico Simoncelli <fsimonce(a)redhat.com> --- M vdsm/storage/volume.py 1 file changed, 3 insertions(+), 3 deletions(-) git pull ssh://gerrit.ovirt.org:29418/vdsm refs/changes/20/26920/1 diff --git a/vdsm/storage/volume.py b/vdsm/storage/volume.py index e1d3fc7..a36914f 100644 --- a/vdsm/storage/volume.py +++ b/vdsm/storage/volume.py @@ -271,7 +271,7 @@ wasleaf = True self.setInternal() try: - self.prepare(rw=False) + self.prepare(rw=False, justme=True) dst_path = os.path.join(dst_image_dir, dst_volUUID) self.log.debug('cloning volume %s to %s', self.volumePath, dst_path) @@ -283,15 +283,15 @@ qemuimg.create(dst_path, backing=parent, format=fmt2str(volFormat), backingFormat=fmt2str(self.getFormat())) - self.teardown(self.sdUUID, self.volUUID) except Exception as e: self.log.exception('cannot clone volume %s to %s', self.volumePath, dst_path) # FIXME: might race with other clones if wasleaf: self.setLeaf() - self.teardown(self.sdUUID, self.volUUID) raise se.CannotCloneVolume(self.volumePath, dst_path, str(e)) + finally: + self.teardown(self.sdUUID, self.volUUID, justme=True) def _shareLease(self, dstImgPath): """ -- To view, visit http://gerrit.ovirt.org/26920 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: newchange Gerrit-Change-Id: Idc009fac4dc1a258537b0ffb15bd627680d79330 Gerrit-PatchSet: 1 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: Federico Simoncelli <fsimonce(a)redhat.com>

3 4

Change in vdsm[master]: task: support task id in client request
by Federico Simoncelli 19 Oct '15

19 Oct '15

Federico Simoncelli has uploaded a new change for review. Change subject: task: support task id in client request ...................................................................... task: support task id in client request Change-Id: Ib5034e6c3466d5a663699d4f924975b7e067c768 Signed-off-by: Federico Simoncelli <fsimonce(a)redhat.com> --- M vdsm/BindingXMLRPC.py M vdsm/clientIF.py M vdsm/storage/dispatcher.py 3 files changed, 16 insertions(+), 1 deletion(-) git pull ssh://gerrit.ovirt.org:29418/vdsm refs/changes/09/26809/1 diff --git a/vdsm/BindingXMLRPC.py b/vdsm/BindingXMLRPC.py index 76251f5..d449088 100644 --- a/vdsm/BindingXMLRPC.py +++ b/vdsm/BindingXMLRPC.py @@ -98,6 +98,7 @@ Create xml-rpc server over http or https. """ HTTP_HEADER_FLOWID = "FlowID" + HTTP_HEADER_TASKID = "TaskID" threadLocal = self.cif.threadLocal @@ -200,6 +201,7 @@ def parse_request(self): r = basehandler.parse_request(self) threadLocal.flowID = self.headers.get(HTTP_HEADER_FLOWID) + threadLocal.taskID = self.headers.get(HTTP_HEADER_TASKID) return r def finish(self): @@ -207,6 +209,7 @@ threadLocal.client = None threadLocal.server = None threadLocal.flowID = None + threadLocal.taskID = None if sys.version_info[:2] == (2, 6): # Override BaseHTTPServer.BaseRequestHandler implementation to @@ -246,6 +249,10 @@ fmt += " flowID [%s]" logargs.append(self.cif.threadLocal.flowID) + if getattr(self.cif.threadLocal, 'taskID', None) is not None: + fmt += " taskID [%s]" + logargs.append(self.cif.threadLocal.taskID) + self.log.debug(fmt, *logargs) try: diff --git a/vdsm/clientIF.py b/vdsm/clientIF.py index eac7950..21e3c71 100644 --- a/vdsm/clientIF.py +++ b/vdsm/clientIF.py @@ -99,6 +99,7 @@ self.channelListener.start() self.threadLocal = threading.local() self.threadLocal.client = '' + self.irs.setClientThreadLocal(self.threadLocal) except: self.log.error('failed to init clientIF, ' 'shutting down storage dispatcher') diff --git a/vdsm/storage/dispatcher.py b/vdsm/storage/dispatcher.py index 6586492..3e2bf0a 100644 --- a/vdsm/storage/dispatcher.py +++ b/vdsm/storage/dispatcher.py @@ -45,10 +45,14 @@ self.storage_repository = config.get('irs', 'repository') self._exposeFunctions(obj) self.log.info("Starting StorageDispatcher...") + self._clientThreadLocal = None @property def ready(self): return getattr(self._obj, 'ready', True) + + def setClientThreadLocal(self, clientThreadLocal): + self._clientThreadLocal = clientThreadLocal def _exposeFunctions(self, obj): for funcName in dir(obj): @@ -66,7 +70,10 @@ @wraps(func) def wrapper(*args, **kwargs): try: - ctask = task.Task(id=None, name=name) + ctaskid = getattr(self._clientThreadLocal, 'taskID', None) + if ctaskid is not None: + self.log.info('using client requested taskID %s', ctaskid) + ctask = task.Task(id=ctaskid, name=name) try: response = self.STATUS_OK.copy() result = ctask.prepare(func, *args, **kwargs) -- To view, visit http://gerrit.ovirt.org/26809 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: newchange Gerrit-Change-Id: Ib5034e6c3466d5a663699d4f924975b7e067c768 Gerrit-PatchSet: 1 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: Federico Simoncelli <fsimonce(a)redhat.com>

4 9

Change in vdsm[master]: sp: prevent master demotion on activation
by Federico Simoncelli 19 Oct '15

19 Oct '15

Federico Simoncelli has uploaded a new change for review. Change subject: sp: prevent master demotion on activation ...................................................................... sp: prevent master demotion on activation Change-Id: I47a0c3ea16bcdefa99899e04c453eaf995344dfb Signed-off-by: Federico Simoncelli <fsimonce(a)redhat.com> --- M vdsm/storage/sp.py 1 file changed, 9 insertions(+), 1 deletion(-) git pull ssh://gerrit.ovirt.org:29418/vdsm refs/changes/32/28332/1 diff --git a/vdsm/storage/sp.py b/vdsm/storage/sp.py index 3f983b6..3762c10 100644 --- a/vdsm/storage/sp.py +++ b/vdsm/storage/sp.py @@ -1041,7 +1041,15 @@ # Domain conversion requires the links to be present self._refreshDomainLinks(dom) - self._backend.setDomainRegularRole(dom) + + # This should never happen because we're not deactivating the + # current master in deactivateStorageDomain if a new master is + # not provided. It is also impossible to connect to a pool + # where the master domain is not active. Anyway to be on the + # safe side we must prevent the current master domain from + # being demoted to regular. + if sdUUID != self.masterDomain.sdUUID: + self._backend.setDomainRegularRole(dom) if dom.getDomainClass() == sd.DATA_DOMAIN: self._convertDomain(dom) -- To view, visit http://gerrit.ovirt.org/28332 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: newchange Gerrit-Change-Id: I47a0c3ea16bcdefa99899e04c453eaf995344dfb Gerrit-PatchSet: 1 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: Federico Simoncelli <fsimonce(a)redhat.com>

5 8

Change in vdsm[master]: [WIP] BZ#844656 Release the lock during _findDomain
by Federico Simoncelli 19 Oct '15

19 Oct '15

Federico Simoncelli has uploaded a new change for review. Change subject: [WIP] BZ#844656 Release the lock during _findDomain ...................................................................... [WIP] BZ#844656 Release the lock during _findDomain Signed-off-by: Federico Simoncelli <fsimonce(a)redhat.com> Change-Id: I8088d5fe716a3a08c3e5cef2d2d9a654ee96f60a --- M vdsm/storage/sdc.py 1 file changed, 21 insertions(+), 7 deletions(-) git pull ssh://gerrit.ovirt.org:29418/vdsm refs/changes/22/6822/1 -- To view, visit http://gerrit.ovirt.org/6822 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: newchange Gerrit-Change-Id: I8088d5fe716a3a08c3e5cef2d2d9a654ee96f60a Gerrit-PatchSet: 1 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: Federico Simoncelli <fsimonce(a)redhat.com>

8 18

Change in vdsm[master]: tests: Add storage mailbox tests
by Nir Soffer 19 Oct '15

19 Oct '15

Nir Soffer has uploaded a new change for review. Change subject: tests: Add storage mailbox tests ...................................................................... tests: Add storage mailbox tests Use fake repository for testing the content of the storage mailbox without using real block devices. Change-Id: If02ed99b95dfd0d6bc5cc9694e60c3808d0974aa Signed-off-by: Nir Soffer <nsoffer(a)redhat.com> --- M tests/Makefile.am A tests/mailboxTests.py 2 files changed, 91 insertions(+), 0 deletions(-) git pull ssh://gerrit.ovirt.org:29418/vdsm refs/changes/71/29371/1 diff --git a/tests/Makefile.am b/tests/Makefile.am index 6507165..a1fd385 100644 --- a/tests/Makefile.am +++ b/tests/Makefile.am @@ -46,6 +46,7 @@ libvirtconnectionTests.py \ lsblkTests.py \ lvmTests.py \ + mailboxTests.py \ main.py \ md_utils_tests.py \ miscTests.py \ diff --git a/tests/mailboxTests.py b/tests/mailboxTests.py new file mode 100644 index 0000000..ec6854c --- /dev/null +++ b/tests/mailboxTests.py @@ -0,0 +1,90 @@ +import ConfigParser +import errno +import os +import time + +from storage import sd +from storage import storage_mailbox +from vdsm import config + +from testrunner import VdsmTestCase as TestCaseBase +import monkeypatch + +# Don't use /tmp, as O_DIRECT does not work with tempfs file system. Keeping +# the test files in the source is ugly but make it very easy to debug the +# mailbox content. +REPO_DIR = 'mailboxTests.tmp' + +HOST_ID = 1 +POOL_ID = 'pool' +MD_DIR = os.path.join(REPO_DIR, POOL_ID, "mastersd", sd.DOMAIN_META_DATA) +INBOX = os.path.join(MD_DIR, "inbox") +OUTBOX = os.path.join(MD_DIR, "outbox") +MAX_HOSTS = 2000 + +# U (0x55) is a nice initial value +DIRTY_MAILBOX = 'U' * storage_mailbox.MAILBOX_SIZE + +fake_config = ConfigParser.ConfigParser() +config.set_defaults(fake_config) +fake_config.set('irs', 'repository', REPO_DIR) + + +class HSMMailboxTests(TestCaseBase): + + def setUp(self): + # Note: we don't remove the inbox and outbox when test ends to make it + # eaiser to debug by checking inbox and outbox content after a test + # fails. + create_repository() + init_mailbox(INBOX) + init_mailbox(OUTBOX) + + @monkeypatch.MonkeyPatch(storage_mailbox, 'config', fake_config) + def test_init_inbox(self): + mailer = storage_mailbox.HSM_Mailbox(HOST_ID, POOL_ID) + try: + time.sleep(0.5) + with open(INBOX) as f: + # First mailbox is not used + data = f.read(storage_mailbox.MAILBOX_SIZE) + self.assertEquals(data, DIRTY_MAILBOX) + # When mailbox is started, it clears the host inbox + data = f.read(storage_mailbox.MAILBOX_SIZE) + self.assertEquals(data, storage_mailbox.EMPTYMAILBOX) + # This mailbox belong to another host, and should not change + data = f.read(storage_mailbox.MAILBOX_SIZE) + self.assertEquals(data, DIRTY_MAILBOX) + finally: + mailer.stop() + + @monkeypatch.MonkeyPatch(storage_mailbox, 'config', fake_config) + def test_keep_outbox(self): + mailer = storage_mailbox.HSM_Mailbox(HOST_ID, POOL_ID) + try: + time.sleep(0.5) + with open(OUTBOX) as f: + # First mailbox is not used + data = f.read(storage_mailbox.MAILBOX_SIZE) + self.assertEquals(data, DIRTY_MAILBOX) + # Outbox is not touched + data = f.read(storage_mailbox.MAILBOX_SIZE) + self.assertEquals(data, DIRTY_MAILBOX) + # This mailbox belong to another host, and should not change + data = f.read(storage_mailbox.MAILBOX_SIZE) + self.assertEquals(data, DIRTY_MAILBOX) + finally: + mailer.stop() + + +def init_mailbox(path): + with open(path, 'w') as f: + f.write(DIRTY_MAILBOX * MAX_HOSTS) + + +def create_repository(): + try: + os.makedirs(MD_DIR) + except OSError as e: + if e.errno != errno.EEXIST: + raise -- To view, visit http://gerrit.ovirt.org/29371 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: newchange Gerrit-Change-Id: If02ed99b95dfd0d6bc5cc9694e60c3808d0974aa Gerrit-PatchSet: 1 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: Nir Soffer <nsoffer(a)redhat.com>

2 3

Change in vdsm[master]: get the status of core dump
by shaohef＠linux.vnet.ibm.com 19 Oct '15

19 Oct '15

ShaoHe Feng has uploaded a new change for review. Change subject: get the status of core dump ...................................................................... get the status of core dump Change-Id: I5d552db4dbd88762950ec5a113a25c13b73319c8 Signed-off-by: ShaoHe Feng <shaohef(a)linux.vnet.ibm.com> --- M vdsm/API.py M vdsm/BindingXMLRPC.py M vdsm/vm.py M vdsm_api/vdsmapi-schema.json M vdsm_cli/vdsClient.py 5 files changed, 36 insertions(+), 0 deletions(-) git pull ssh://gerrit.ovirt.org:29418/vdsm refs/changes/31/11131/1 diff --git a/vdsm/API.py b/vdsm/API.py index c5f7d40..6b4663a 100644 --- a/vdsm/API.py +++ b/vdsm/API.py @@ -302,6 +302,15 @@ return errCode['noVM'] return v.dumpCancel() + def dumpStatus(self): + """ + Report status of a currently outgoing core dump process. + """ + v = self._cif.vmContainer.get(self._UUID) + if not v: + return errCode['noVM'] + return v.dumpStatus() + def desktopLock(self): """ Lock user session in guest operating system using guest agent. diff --git a/vdsm/BindingXMLRPC.py b/vdsm/BindingXMLRPC.py index 17d97b1..b1f22fd 100644 --- a/vdsm/BindingXMLRPC.py +++ b/vdsm/BindingXMLRPC.py @@ -219,6 +219,10 @@ vm = API.VM(vmId) return vm.dumpCancel() + def vmCoreDumpStatus(self, vmId): + vm = API.VM(vmId) + return vm.dumpStatus() + def vmReset(self, vmId): vm = API.VM(vmId) return vm.reset() @@ -769,6 +773,7 @@ (self.vmCont, 'cont'), (self.vmCoreDump, 'coreDump'), (self.vmCoreDumpCancel, 'dumpCancel'), + (self.vmCoreDumpStatus, 'dumpStatus'), (self.vmSnapshot, 'snapshot'), (self.vmMerge, 'merge'), (self.vmMergeStatus, 'mergeStatus'), diff --git a/vdsm/vm.py b/vdsm/vm.py index 0a40e97..5d9c0d9 100644 --- a/vdsm/vm.py +++ b/vdsm/vm.py @@ -1371,3 +1371,6 @@ return reportError(msg=e.message) finally: self._guestCpuLock.release() + + def dumpStatus(self): + return self._doCoredumpThread.getStat() diff --git a/vdsm_api/vdsmapi-schema.json b/vdsm_api/vdsmapi-schema.json index 39d1cba..e96f01f 100644 --- a/vdsm_api/vdsmapi-schema.json +++ b/vdsm_api/vdsmapi-schema.json @@ -5484,6 +5484,16 @@ {'command': {'class': 'VM', 'name': 'dumpCancel'}} ## +# @VM.dumpStatus: +# +# Reports the state of the currently core dump process +# +# Since: 4.10.4 +# +## +{'command': {'class': 'VM', 'name': 'dumpStatus'}} + +## # @VM.monitorCommand: # # Send a command to the qemu monitor. diff --git a/vdsm_cli/vdsClient.py b/vdsm_cli/vdsClient.py index 32ad348..7edc674 100644 --- a/vdsm_cli/vdsClient.py +++ b/vdsm_cli/vdsClient.py @@ -1674,6 +1674,11 @@ response = self.s.dumpCancel(vmId) return response['status']['code'], response['status']['message'] + def do_dumpStat(self, args): + vmId = args[0] + response = self.s.dumpStatus(vmId) + return response['status']['code'], response['status']['message'] + def coreDump(self, args): dumpParams = {'crash': False, 'live': False, @@ -2422,6 +2427,10 @@ ('<vmId>', 'cancel machine core dump' )), + 'coreDumpStatus': (serv.do_dumpStat, + ('<vmId>', + 'Check the progress of current core dump' + )), 'coreDump': (serv.coreDump, ('<vmId> <file> [live=<True|False>] ' '[crash=<True|False>] [bypass-cache=<True|False>] ' -- To view, visit http://gerrit.ovirt.org/11131 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: newchange Gerrit-Change-Id: I5d552db4dbd88762950ec5a113a25c13b73319c8 Gerrit-PatchSet: 1 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: ShaoHe Feng <shaohef(a)linux.vnet.ibm.com>

10 59

Change in vdsm[master]: dump the core of a domain
by shaohef＠linux.vnet.ibm.com 19 Oct '15

19 Oct '15

ShaoHe Feng has uploaded a new change for review. Change subject: dump the core of a domain ...................................................................... dump the core of a domain libvirt support an API to dump the core of a domain on a given file for analysis when guest OS crash. There are two kind of dump files. one is QEMU suspend to disk image. the other is core file which like kdump file butcontains registers' value. It's helpful to support by VDSM to find root cause if a guest gets hang and kdump isn't set up in it. This would be a good RAS feature. Here's the definition of the new API: coreDump: This method will dump the core of a domain on a given file for analysis. Input parameter: vmId - VM UUID to - the core file named by the user flags - defined in libvirt.py VIR_DUMP_CRASH VIR_DUMP_LIVE VIR_DUMP_BYPASS_CACHE VIR_DUMP_RESET VIR_DUMP_MEMORY_ONLY Return value: success: return doneCode failure: return errCode including underlying libvirt error message. Change-Id: If4aac9e747dc7aa64a6ff5ef256a7a4375aa2bb5 Signed-off-by: ShaoHe Feng <shaohef(a)linux.vnet.ibm.com> --- M vdsm/API.py M vdsm/BindingXMLRPC.py M vdsm/define.py M vdsm/libvirtvm.py M vdsm_cli/vdsClient.py 5 files changed, 80 insertions(+), 0 deletions(-) git pull ssh://gerrit.ovirt.org:29418/vdsm refs/changes/29/7329/1 diff --git a/vdsm/API.py b/vdsm/API.py index 19cbb42..e2b24cb 100644 --- a/vdsm/API.py +++ b/vdsm/API.py @@ -244,6 +244,12 @@ self.log.debug("Error creating VM", exc_info=True) return errCode['unexpected'] + def coreDump(self, to, flags): + v = self._cif.vmContainer.get(self._UUID) + if not v: + return errCode['noVM'] + return v.coreDump(to, flags) + def desktopLock(self): """ Lock user session in guest operating system using guest agent. diff --git a/vdsm/BindingXMLRPC.py b/vdsm/BindingXMLRPC.py index cc5300f..be71e6a 100644 --- a/vdsm/BindingXMLRPC.py +++ b/vdsm/BindingXMLRPC.py @@ -208,6 +208,10 @@ vm = API.VM(vmId) return vm.cont() + def vmCoreDump(self, vmId, to, flags): + vm = API.VM(vmId) + return vm.coreDump(to, flags) + def vmReset(self, vmId): vm = API.VM(vmId) return vm.reset() @@ -725,6 +729,7 @@ (self.getVMList, 'list'), (self.vmPause, 'pause'), (self.vmCont, 'cont'), + (self.vmCoreDump, 'coreDump'), (self.vmSnapshot, 'snapshot'), (self.vmMerge, 'merge'), (self.vmMergeStatus, 'mergeStatus'), diff --git a/vdsm/define.py b/vdsm/define.py index 31deb4f..1fedac5 100644 --- a/vdsm/define.py +++ b/vdsm/define.py @@ -114,6 +114,10 @@ 'mergeErr': {'status': {'code': 52, 'message': 'Merge failed'}}, + 'coreDumpErr': {'status': + {'code': 54, + 'message': + 'Failed to get coreDump file'}}, 'recovery': {'status': {'code': 99, 'message': diff --git a/vdsm/libvirtvm.py b/vdsm/libvirtvm.py index 4554fee..cbd9f96 100644 --- a/vdsm/libvirtvm.py +++ b/vdsm/libvirtvm.py @@ -1904,6 +1904,27 @@ self.saveState() + def coreDump(self, to, flags): + + def reportError(key='coreDumpErr', msg=None): + self.log.error("get coreDump failed", exc_info=True) + if msg == None: + error = errCode[key] + else: + error = {'status' : {'code': errCode[key] \ + ['status']['code'], 'message': msg}} + return error + + if self._dom == None: + return reportError() + try: + self._dom.coreDump(to, flags) + except libvirt.libvirtError, e: + if e.get_error_code() == libvirt.VIR_ERR_NO_DOMAIN: + return reportError(key='noVM') + return reportError(msg=e.message) + return {'status': doneCode} + def changeCD(self, drivespec): return self._changeBlockDev('cdrom', 'hdc', drivespec) diff --git a/vdsm_cli/vdsClient.py b/vdsm_cli/vdsClient.py index eeb7c95..cdcd3a8 100644 --- a/vdsm_cli/vdsClient.py +++ b/vdsm_cli/vdsClient.py @@ -1589,6 +1589,33 @@ return status['status']['code'], status['status']['message'] + def coreDump(self, args): + DUMPFLAGS = {'crash': 1 << 0, + 'live': 1 << 1, + 'bypass-cache': 1 << 2, + 'reset': 1 << 3, + 'memory-only': 1 << 4} + flags = 0 + vmId = args[0] + coreFile = args[1] + params = {} + if len(args) > 2: + for arg in args[2:]: + kv = arg.split('=', 1) + if len(kv) < 2: + params[kv[0]] = "True" + else: + params[kv[0]] = kv[1] + for k, v in params.items(): + if v.lower() == "true" or not v: + try: + flags = flags + DUMPFLAGS[k] + except KeyError: + print "unrecognized optoin %s for cormDump command" % k + response = self.s.coreDump(vmId, coreFile, flags) + return response['status']['code'], response['status']['message'] + + if __name__ == '__main__': if _glusterEnabled: serv = ge.GlusterService() @@ -2239,6 +2266,23 @@ ('<vmId> <sdUUID> <imgUUID> <baseVolUUID> <volUUID>', "Take a live snapshot" )), + 'coreDump': (serv.coreDump, + ('<vmId> <file> [live=<True>] ' + '[crash=<True>] [bypass-cache=<True>] ' + '[reset=<True>] [memory-only=<True>]', + "get memeory dump or migration file" + 'optional params:', + 'crash: crash the domain after core dump' + 'default False', + 'live: perform a live core dump if supported, ' + 'default False', + 'bypass-cache: avoid file system cache when saving' + 'default False', + 'reset: reset the domain after core dump' + 'default False', + "memory-only: dump domain's memory only" + 'default False' + )), } if _glusterEnabled: commands.update(ge.getGlusterCmdDict(serv)) -- To view, visit http://gerrit.ovirt.org/7329 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: newchange Gerrit-Change-Id: If4aac9e747dc7aa64a6ff5ef256a7a4375aa2bb5 Gerrit-PatchSet: 1 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: ShaoHe Feng <shaohef(a)linux.vnet.ibm.com>

16 96

← Newer
1
...
6
7
8
9
10
11
12
...
84
Older →

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

vdsm-patches May 2015