Add simple lock monitor
This patch adds an initial implementation of a lock monitor, accessiblefor the user through “gnt-debug locks”. It currently shows all resourcelocks: BGL, nodes and instances. Config and job queue locks could beshown too, but wouldn't be of much help. The current owner(s) and mode...
workerpool: Allow setting task name
With this patch, the task name is added to the thread name and will show up inlogs. Log messages from jobs will look like “pid=578/JobQueue14/Job13 mcpu:289DEBUG LU locks acquired/cluster/BGL/shared”.
Signed-off-by: Michael Hanselmann <hansmi@google.com>...
Use one function to parse “--fields” option values
Signed-off-by: Michael Hanselmann <hansmi@google.com>Reviewed-by: Iustin Pop <iustin@google.com>
locking.LockSet: Use function to get member lock name
Finalize removal of “lock_status” job field
When removing the field in commit 9bdab62198, I forgot to remove some parts.
Prepare 2.2.0~rc1 release
Update NEWS and version number.
setup-ssh: fix updating of authorized_keys
Due to what seems like a bug (or inconsistency) in paramiko, filesopened with a+ over SFTP need a seek() in order for the user to be ableto read data from them. We implement this, and rely on the fact that wedo iterate over all lines before writing and that the file is opened in...
setup-ssh: Also use keys from the ssh-agent
Currently, setup-ssh only uses one disk-based key. This means that anysetup where we use keys from ssh-agent (which do not necessarily existon disk) will break when moving from the old method to setup-ssh.
This patch moves the SSH key handling to separate functions, and uses...
setup-ssh: fix the logging error message
This took a bit of interesting debugging, but the issue seems to be thaton older python/logging/paramiko (I'm not sure which is at fault here),the other threads are finishing later than the main thread, so thelogging module closes the logging file (but doesn't unregister it)...
setup-ssh: only read the ssh port once
Then reuse it. With certain setups, this could be slow when pre-seedingan entire cluster.
Signed-off-by: Iustin Pop <iustin@google.com>Reviewed-by: Guido Trotter <ultrotter@google.com>Reviewed-by: René Nussbaumer <rn@google.com>
setup-ssh: redo the logging levels
There is not enough logged by default on stderr, and way too muchinformation in the log file.
Since we don't want to debug paramiko itself, we filter only warningsand above.
Otherwise, we make WARNINGS the default level, with verbose meaning INFO...
setup-ssh: try to use key auth first
This patch changes the setup-ssh workflow to try key authenticationfirst, and then fall-back to password authentication. The password isalso read lazily, with no prompts if we can authenticate via keys.
Signed-off-by: Iustin Pop <iustin@google.com>...
Use Sphinx' :rfc: extension to refer to RFCs
Document non-standard usage of JSON in RAPI
This was requested in issue 118.
Fix small spelling mistake
Signed-off-by: Manuel Franceschini <livewire@google.com>Reviewed-by: Iustin Pop <iustin@google.com>
Fix a few commands behaviour with dry-run
These commands use or display the result from the LU, so in case ofdry-run, they will crash or display just 'None'. At least checking thatthe result is 'true' (in the boolean sense) will make them work better....
Stop adding the dry-run option by default
Currently cli.py unconditionally adds the dry-run option. This patchdisables this, and exports dry-run as a normal option.
The other alternative I tried to implement (adding a new fake option fordisabling the auto-add per individual command) would require changes in...
Explicitly add dry-run to some commands
Based on manual inspection (that the command only does a submit of somejobs/opcodes), we re-add the dry-run option on a subset of the existingcommands.
A few more commands could use dry-run, but the code doesn't deal nicely...
jqueue: Remove lock status field
With the job queue changes for Ganeti 2.2, watched and queried jobs areloaded directly from disk, rendering the in-memory “lock_status” fielduseless. Writing it to disk would be possible, but has a huge cost atruntime (when tested, processing 1'000 opcodes involved 4'000 additional...
QA: Run simple job queue test
Merge branch 'devel-2.1' into devel-2.2
Signed-off-by: Guido Trotter <ultrotter@google.com>Reviewed-by: Michael Hanselmann <hansmi@google.com>
Don't ignore secondary node silently
Currently on non-mirrored disk templates the secondary node is ignoredsilently. This patch adds a check for this case, and warns the usershould this be happening. This solves issue 113.
The patch also moves a prereq check to an argument check. This is ok...
etags: force Python as a language
etags has cannot properly identify the scripts/gnt-* files as Python, sowe statically force the language to Python.
Signed-off-by: Iustin Pop <iustin@google.com>Reviewed-by: Michael Hanselmann <hansmi@google.com>
Conflicts: man/gnt-instance.sgml...
RAPI client: Support modifying instances
RAPI: Allow modifying instance
Small fixes for instance creation via RAPI documentation
- Inconsistencies- Missing types
gnt-debug: Extend job queue tests
Test multiple opcodes, also with failure.
jqueue: Mark opcodes following failed ones as failed, too
When an opcode fails, the job queue would leave following opcodes as “queued”,which can be quite confusing. With this patch, they're all marked as failed andassertions are added to check this.
jqueue: Work around race condition between job processing and archival
This is a simplified version of a patch I sent earlier to make sure the jobfile is only written once with a finalized status.
jqueue: More checks for cancelling queued job
We can also check when the lock status is updated. This willimprove job cancelling.
errors: Function to check whether value is encoded error
jqueue: Add more debug output
gnt-backup: Pass error code to OpPrereqError
Merge branch 'devel-2.1'
Re-add the 'live' parameter to migration opcodes
This patch reintroduces the live parameter, for backwards compatibilityat the Luxi level. This way, clients can work transparently with both2.1 and 2.2, even though sub-optimally.
Fix --master-netdev arg name in gnt-cluster(8)
This fixes Issue 114.
Restore 'tablet mouse on vnc' behavior
We needed to revert commit 5b062a58ac76b39c2dc6a7e1543affdf43dc7ee7because it was in conflict with the usb_mouse hv parameter. Here wereintroduce its functionality only when usb_mouse is not specified.
Signed-off-by: Guido Trotter <ultrotter@google.com>...
Document the usb_mouse hv parameter
Revert "Add -usbdevice tablet to KVM when using vnc"
This reverts commit 5b062a58ac76b39c2dc6a7e1543affdf43dc7ee7.This fixes issue 109. The mouse type can be set with the usb_mousehv parameter.
Fix path in ganeti-rapi man page
This takes care of issue 116.
workerpool: Don't keep reference to task arguments
The workerpool should not keep any reference to task arguments afterthey were processed by RunTask. Doing so led to jobs being cachedby the job queue's WeakValueDictionary for longer than they should've...
cli.SubmitOpCode: Pass keyword parameter as keyword
Signed-off-by: Michael Hanselmann <hansmi@google.com>Reviewed-by: Manuel Franceschini <livewire@google.com>
gnt-backup: Don't show confusing message w/o target node
“gnt-backup export” requires the target node. Until now, the masterdaemon would complain that the “parameter'OP_BACKUP_EXPORT.target_node' fails validation”. With this patch,an additional check is done in the client program....
masterd.instance: Add missing argument
_DiskTransferPrivate takes three parameters, not two.
Adjust message in case ~/.ssh is no directory
Use actual path, not something hardcoded.
Signed-off-by: Michael Hanselmann <hansmi@google.com>Reviewed-by: Guido Trotter <ultrotter@google.com>
RAPI client: Fix docstring for migrating instance
QA: Test renaming instance via RAPI
RAPI client: Support renaming instances
Allow renaming instances via RAPI
RAPI client: Don't re-use PycURL object
With this patch, a new PycURL object will be created for each request.This should make the RAPI client safe for simultaneous calls frommultiple threads. Unittests are adjusted accordingly.
An unnecessary variable assignment is also removed from the unittest...
Add --no-name-check to 'gnt-instance rename' man page
Signed-off-by: Manuel Franceschini <livewire@google.com>Reviewed-by: Michael Hanselmann <hansmi@google.com>
Fix bug in bdev when drbd version format is x.x.x.x
This patch fixes a bug reported in [0]. Newer drbd versions can haveanother digit beyond the regular major, minor and point release digits.We modify the regex used to match that with an optional parts which is...
Bump version to 2.2.0~rc0
move-instance: Use constants for parameters
Allow instance NIC's IP address to be None
Also add some assertions.
Test instance NIC and disk parameter names
Add new parameter type “maybe string”
Before strict checking was implemented, NIC IP addresses could be setto “None”. Commit bd061c35 added more strict checking, includingenforcing the IP address to be a string. With this new type, itcan again be set to None....
cmdlib: Change expected type for source CA on remote import
Signed-off-by: Michael Hanselmann <hansmi@google.com>Reviewed-by: Guido Trotter <ultrotter@google.com>Reviewed-by: Iustin Pop <iustin@google.com>
move-instance: Pass OS parameters to new instance
Update NEWS file for the first release candidate
Fix a few job archival issues
This patch fixes two issues with job archival. First, theLoadJobFromDisk can return 'None' for no-such-job, and we shouldn't addNone to the job list; we can't anyway, as this raises an exception:
node1# gnt-job archive foo...
burning: fix handling of empty job sets
If we call burning with only existing instance, then it will fail tocreate any of them, and thus in the removal phase it won't have anythingto remove. Since calling luxi.SUBMIT_MULTIPLE_JOBS with an empty job set...
Change semantics of --force-multi for reinstall
Currently, we require both --force and --force-multiple for skipping theconfirmation on instance reinstalls. After offline conversations, thishas been deemed to be excessive, and this patch changes the meaning of...
Change handling of non-Ganeti errors in jqueue
Currently, if a job execution raises a Ganeti-specific error (i.e.subclass of GenericError), then we encode it as (error class, [errorargs]). This matches the RAPI documentation.
However, if we get a non-Ganeti error, then we encode it as simply...
Implement gnt-cluster master-ping
This can be used from shell-scripts to quickly check the status of themaster node, before launching a series of jobs (and handling the failureof the jobs due to masterd other issues).
Instance migration: remove error on missing link
Since we don't support upgrades from 1.2.4 without restarting theinstance, the 'not restarted since 1.2.5' check/error iswrong/misleading.
Since the live migration works anyway without the links (it recreates...
Add check for RAPI paths to start with /2
During a discussion in July 2010 it was decided that we'll stabilize on /2. Seemessage ID <20100716180012.GA9423@google.com> for reference.
Ensure assertions are evaluated in tests
A lot of assertions are used in Ganeti's code. Some unittests even checkwhether AssertionError is raised in some cases. Explicitely ensuringassertions are evaluated makes sure those tests don't fail andassertions are checked....
RAPI client: The os argument for instance reinstalls is optional
Signed-off-by: David Knowles <dknowles@google.com>Reviewed-by: Michael Hanselmann <hansmi@google.com>
RAPI client: Support migrating instances
QA: Test instance migration via CLI and RAPI
workerpool: Add two additional assertions
workerpool: Change signature of AddTask function to not use *args
By changing it to a normal parameter, which must be a sequence, we canstart using keyword parameters.
Before this patch all arguments to “AddTask(self, *args)” were passed asarguments to the worker's “RunTask” method. Priorities, which should be...
RAPI: Support migrating instances
workerpool: Move waiting for new tasks for a worker to the pool
This way fewer private variables of the pool are accesssed by the worker.
workerpool: Remove unused worker method
HasRunningTask is never used except for an assertion, where wedon't really need the lock.
workerpool: Additional check in BaseWorker.ShouldTerminate
Document that it should only be called from within RunTask andadd an assertion for this. This means we can no longer use amethod on the pool and hence remove WorkerPool.ShouldWorkerTerminate.
workerpool: Use common function to add tasks
Fix install document regarding DRBD usage
This is related to issue 105.
Update RAPI documentation for the OS changes
Rename masterfailover to master-failover
Most (all?) of our commands use dash-separator: replace-disks,verify-disks, add-tags, etc. “gnt-cluster masterfailover” is an oldexception to this rule.
The patch replaces it with master-failover, add a compatiblity alias,...
RAPI: Add os params to instance creation v1
Since the RAPI QA suite doesn't seem to offer easy testing of failedcreations, I didn't add this to the QA. Pointers on how to do it arewelcome.
The patch also changes the 'os' argument to be required, since that is...
makefile: fix TAGS building
“find .” requires that “-path” arguments start with a dot, otherwisethey are not matches. Additionally, we also include the QA files in thetags, for easier search while modifying the QA suite.
Improve handling of lost jobs
Currently, if the cli.JobExecutor class is being used, and one of thejobs is being archived before it can check its result, it will raise astracktrace as _ChooseJob is not prepared to handle this case.
This case makes JobExecutor work better with lost jobs (it still reports...
luxi: convert permission errors into exception
This patch adds handling of permission errors so that we don't showtracebacks when a non-root user runs a gnt-* command. Since in thefuture we'll have different permissions, we need to handle this in RAPItoo....
cmdlib: Return new name from rename operations
The new name is then displayed by the clients.
gnt-instance rename: Fix bug and rename params
This patch fixes a bug when gnt-instance rename was invoked with--no-name-check. It renames the internal variables to be consistent withthe ones in equivalent instance add code. Furthermore it checks whether...
QA: add tests for the reserved lvs feature
Signed-off-by: Iustin Pop <iustin@google.com>Reviewed-by: René Nussbaumer <rn@google.com>
Add modification of the reserved logical volumes
This doesn't allow addition/removal of individual volumes, onlywholesale replace of the entire list. It can be improved later, if weever get generic container parameters.
The man page changes replaces some tabs with spaces (hence the...
Add printing of reserved_lvs in cluster info
Introuce a new cluster parameter - reserved_lvs
This parameter, which is a list of regular expression patterns, willmake cluster verify ignore any such LVs. It will not prevent creation orremoval of such volumes by the backend code.
Change the meaning of call_node_start_master
Currently, backend.StartMaster (the function behind this RPC call) willactivate the master IP and then, if the start_daemons parameter is true,it will also activate the master role.
While this works, it has two issues:...
masterd: move the IP activation from Exec to Check
Currently, the master IP activation is done in the Exec function. Sincethe original masterd process returns after forking, and Exec is run inthe (grand)child process, this means that after 'ganeti-masterd' has...
Move the UsesRPC decorator from cli to rpc
This is needed because not just the cli scripts need this decorator, butthe master daemon too (and it already duplicated the code once).
In cli.py we just leave a stub, so that we don't have to modify all thescripts to import rpc.py....
watcher: smarter handling of instance records
This patch implements a few changes to the instance handling. First, oldinstances which no longer exist on the cluster are removed from thestate file, to keep things clean.
Second, the instance restart counters are reset every 8 hours, since...
Update the RAPI node migrate for the 'live' change
This patch adds handling of the new 'mode' parameter to the RAPI server,while keeping compatibility with the old mode. Note that in the old mode(when 'live' is being passed), the auto-mode doesn't work....
Update the RAPI client for the migration mode
See the discussion on the previous patch about this. Basically unless wewant to a add a new 'feature' marking for the live migration parameter,there is no simple way to handle this nicely in the client.
Given that the client was/is marked as experimental, this patch simply...
Fix burnin and live migration
This is breakage from the original 'live' parameter changes.
Rename the OpMigrate* parameter 'live' to 'mode'
This is needed as now the parameter is no longer boolean, but tri-state.