Re: On string data types

2016-06-07 13:10 GMT+02:00 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com>: > Hi Iustin, > > I'm half inclined to think that if we're going to change the type of UUID > we should do it 'properly', ie change to a strict, unpacked 16-byte array > that implements Eq, Show, Read (and

[PATCH master] Fixup compatibility with GHC 7.4/base 4.5

It looks like commit c429dd26 introduced the use of atomicModifyIORef', which is only present in base 4.6 (GHC 7.6). Let's temporarily fix that by adding a small compat layer (which undoes the optimisations of using strict on 7.4, but at least it works), until we decide to officially drop support

[PATCH stable-2.15] Fixup compatibility with GHC 7.4/base 4.5

It looks like commit c429dd26 introduced the use of atomicModifyIORef', which is only present in base 4.6 (GHC 7.6). Let's fix that by importing the actual implementation of said function from base 4.6 in case we're running with earlier versions. Signed-off-by: Iustin Pop ---

Re: [PATCH master] Fixup compatibility with GHC 7.4/base 4.5

On Fri, Jun 10, 2016 at 12:07:45PM +0100, Brian Foley wrote: > On Fri, Jun 10, 2016 at 12:33:23PM +0200, 'Iustin Pop' via ganeti-devel wrote: > > On Fri, Jun 10, 2016 at 12:30:03PM +0200, Iustin Pop wrote: > > > It looks like commit c429dd26 introduced the use of atomicModifyIO

Re: [PATCH master] Fixup compatibility with GHC 7.4/base 4.5

On Fri, Jun 10, 2016 at 12:30:03PM +0200, Iustin Pop wrote: > It looks like commit c429dd26 introduced the use of atomicModifyIORef', which > is only present in base 4.6 (GHC 7.6). Let's temporarily fix that by adding a > small compat layer (which undoes the optimisations of using strict on 7.4,

Re: [PATCH master 4/8] Split getInstance into *ByUuid and *ByPartialName

On Fri, Jun 10, 2016 at 11:28:16AM +0100, Brian Foley wrote: > On Thu, Jun 09, 2016 at 11:13:54PM +0200, Iustin Pop wrote: > > From: Iustin Pop > > > > This function (and getNode, next patch) were two pain points when I tried to > > convert UUIDs to another type, since they're

Re: [MERGE] stable-2.15 to stable 2.16

2016-06-10 15:28 GMT+02:00 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com>: > commit 5785f214a9e728465a4bfc1aef7ded306225cfa2 > Merge: 40cd52f 2429235 > Author: Brian Foley > Date: Fri Jun 10 14:23:10 2016 +0100 > > Merge branch 'stable-2.15' into

Re: [PATCH master 3/3] Fix KVM pinning tests to not depend on the local machine

2016-06-10 10:58 GMT+02:00 Viktor Bachraty : > > On Thu, Jun 9, 2016 at 11:33 PM, Iustin Pop wrote: > >> From: Iustin Pop >> >> Commit 8b2ec2f added unittests for KVM pinning, but it introduced a >> non-obvious >> local dependency in

Re: [MERGE] stable-2.17 to master

2016-06-10 16:25 GMT+02:00 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com>: > commit 7eb49311e18865db76c4e8da5eb4b2e166db2d55 > Merge: a32eb3c 17a1c27 > Author: Brian Foley > Date: Fri Jun 10 15:20:33 2016 +0100 > > Merge branch 'stable-2.17' > > *

Re: [MERGE] stable-2.16 to stable-2.17

2016-06-10 15:40 GMT+02:00 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com>: > commit b462d8c77bff0789e8a951288dea34226ab8b6d7 > Merge: 20c24a8 90281b4 > Author: Brian Foley > Date: Fri Jun 10 14:35:13 2016 +0100 > > Merge branch 'stable-2.16' into

Re: [PATCH stable-2.16 1/2] Tune getNodeInstances DRBD secondary computation

2016-05-30 11:34 GMT+02:00 Brian Foley : > On Sun, May 29, 2016 at 12:01:39PM +0200, Iustin Pop wrote: > > On Thu, May 26, 2016 at 03:56:10PM +0100, Ganeti Development List wrote: > > > This is exercised by the luxi QueryInstances call when the sinst_cnt > and > > > sint_list

Re: [PATCH instance-debootstrap 0/4] Cleanups and simple tests

2016-05-30 11:51 GMT+02:00 Brian Foley : > On Sun, May 29, 2016 at 02:38:22AM +0200, Iustin Pop wrote: > > From: Iustin Pop > > > > A few cleanups, and adding a test suite to be run without Ganeti, i.e. on > > standalone systems. However it has downsides,

UUID references (was: Tune getNodeInstances DRBD secondary) computation

On Thu, May 26, 2016 at 03:56:10PM +0100, Ganeti Development List wrote: > This is exercised by the luxi QueryInstances call when the sinst_cnt and > sint_list fields are used. This uses a lot of CPU and does a lot of > short-lived heap allocation on clusters with many instances. > > The

Re: UUID references (was: Tune getNodeInstances DRBD secondary) computation

nvasive it gets and I'll send it against stable-2.16 (if not too invasive) or master otherwise. iustin > On Mon, May 30, 2016 at 11:40 AM, 'Iustin Pop' via ganeti-devel < > ganeti-devel@googlegroups.com> wrote: > > > On Thu, May 26, 2016 at 03:56:10PM +0100, Ganeti Development Li

Re: UUID references (was: Tune getNodeInstances DRBD secondary) computation

On Mon, May 30, 2016 at 12:18:37PM +0100, Brian Foley wrote: > On Mon, May 30, 2016 at 12:40:49PM +0200, Iustin Pop wrote: > > On Thu, May 26, 2016 at 03:56:10PM +0100, Ganeti Development List wrote: > > > This is exercised by the luxi QueryInstances call when the sinst_cnt and > > > sint_list

Re: [PATCH stable-2.16 1/3] Fix optimisation: Correctly extract secondary node.

On Mon, May 30, 2016 at 03:47:40PM +0100, Brian Foley wrote: > On Mon, May 30, 2016 at 03:33:50PM +0200, Iustin Pop wrote: > > On Mon, May 30, 2016 at 02:10:14PM +0100, Ganeti Development List wrote: > > > It is possible for either nodeA or nodeB in a DRBD disk to be the > > > primary or the

Re: [PATCH stable-2.16 1/3] Fix optimisation: Correctly extract secondary node.

On Mon, May 30, 2016 at 02:10:14PM +0100, Ganeti Development List wrote: > It is possible for either nodeA or nodeB in a DRBD disk to be the > primary or the secondary. Patch 9825767ad incorrectly assumed nodeB > was the secondary. Correct this by passing the actual primary to >

Re: [PATCH stable-2.16 1/2] Tune getNodeInstances DRBD secondary computation

On Thu, May 26, 2016 at 03:56:10PM +0100, Ganeti Development List wrote: > This is exercised by the luxi QueryInstances call when the sinst_cnt and > sint_list fields are used. This uses a lot of CPU and does a lot of > short-lived heap allocation on clusters with many instances. > > The

[PATCH instance-debootstrap] Fix sfdisk invocation for util-linux 2.26+

sfdisk changed in 2.26; it stopped supporting the geometry options (-C, -H, -S), deprecated -L and -u. However, we can't just drop these options, as pre-2.26 needs them for devices that don't report geometry (e.g. plain files). This patch changes the invocation of sfdisk based on what the

Re: [PATCH instance-debootstrap] Fix sfdisk invocation for util-linux 2.26+

2016-05-26 12:37 GMT+02:00 Brian Foley <bpfo...@google.com>: > On Thu, May 26, 2016 at 12:26:28PM +0200, 'Iustin Pop' via ganeti-devel > wrote: > > sfdisk changed in 2.26; it stopped supporting the geometry options (-C, > -H, > > -S), deprecated -L and -u.

Re: [PATCH stable-2.15 1/2] KVM: handle gracefully too old psutil versions

On 21 June 2016 at 16:28, Brian Foley wrote: > On Sat, Jun 18, 2016 at 04:53:25AM +0200, Iustin Pop wrote: > > On 2016-06-15 10:23:57, Brian Foley wrote: > > > Additionally, 0.5.0 had a psutil.Process.{get,set}_cpu_affinity() API, > > > which we use in the kvm code. In 2.0.0

Re: [PATCH stable-2.15] KVM: handle gracefully too old/too new psutil versions

On 21 June 2016 at 16:29, Brian Foley wrote: > On Sat, Jun 18, 2016 at 04:54:51AM +0200, Iustin Pop wrote: > > From: Iustin Pop > > > > My previous pylint cleanups were done without psutil installed; as soon > > I installed it, pylint showed that the

Re: [PATCH stable-2.17] Prevent watcher from holding lock on verify group

On 17 June 2016 at 10:36, Federico Pareschi wrote: > Yes, this is definitely the case. > We have ideas to implement something in the future to obviate this > problem, although we're still considering exactly how to implement it. > This is a short-term quick fix to solve some

Re: [PATCH stable-2.17] Prevent watcher from holding lock on verify group

2016-06-17 9:46 GMT-07:00 Federico Pareschi : > When a ganeti-watcher runs on the nodegroup, it submits the verify-group > job. If there is another job in the queue that is taking some locks that > stop the verify-group job (like an instance creation) then the whole >

Re: [PATCH master 0/6] New balancing options implementation

, I just wanted to know if Daniil acknowledged the comments or not. Anyway, comments are welcome here and the discussion is still open:) > Sounds good. And thanks Daniil for the commits. > Of course, looking forward to see this implemented! thanks! iustin On Thu, Jun 23, 2016 at 5:49 PM '

Re: [PATCH master 0/6] New balancing options implementation

On 23 June 2016 at 17:42, Даниил Лещёв wrote: > Hi, Iustin > > >> Oh, no worries, I just wanted to know if Daniil acknowledged the comments >> or not. >> >> Anyway, comments are welcome here and the discussion is still open:) >>> >> >> > The only reason why I didn't reply to

Re: [PATCH master 0/6] New balancing options implementation

On 23 June 2016 at 16:45, wrote: > From: Daniil Leshchev > > The patchset introduces new command line options > (--long-solution-threshold" and --avoid-long-solutions"). > That gives an ability for HBal to avoid balancing solutions, > that take significant

Re: [PATCH master 0/6] New balancing options implementation

On 23 June 2016 at 18:32, Даниил Лещёв wrote: > >> I would slightly prefer if we discuss it over plain email (without >> patches), to see what you think about how complex the network model needs >> to be, and whether a static "time X" vs. semi-dynamic (based on the >> instance

Re: [PATCH stable-2.17] Prevent watcher from holding lock on verify group

2016-06-17 8:31 GMT-07:00 'Federico Morg Pareschi' via ganeti-devel < ganeti-devel@googlegroups.com>: > The ganeti-watcher holds the group file lock for too long, until after > the execution of a group-verify-disk job. This locks for a long time if > there are other jobs already running and

Re: [PATCH stable-2.16] Fix typos in gnt-cluster man page

On Fri, Aug 12, 2016 at 03:03:18PM +0100, Ganeti Development List wrote: > Luckily, nothing that changes the meaning anywhere. > > Signed-off-by: Brian Foley LGTM, thanks.

Re: [PATCH stable-2.16] Hide errors for expected inotify failures in unittest

On 12 August 2016 at 15:04, 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com> wrote: > This makes it a little easier to eyeball the output of make py-tests. > Ooh, nice, this is a very old bug, thanks! +logger = logging.getLogger('pyinotify') > +logger.propagate = False >

Re: [PATCH stable-2.16] Hide errors for expected inotify failures in unittest

On 12 August 2016 at 15:38, Brian Foley wrote: > On Fri, Aug 12, 2016 at 03:13:53PM +0200, Iustin Pop wrote: > >On 12 August 2016 at 15:04, 'Brian Foley' via ganeti-devel > ><[1]ganeti-devel@googlegroups.com> wrote: > > > > This makes it a little easier to

Re: [PATCH master 2/6] Add bandwidth tags and bandwidth map fields into Node

On 28 June 2016 at 11:28, Даниил Лещёв wrote: > > > The bandwidths map store data about network speed >> > between current node and given Node by it's bandwidth tags. >> >> Filling this up will take some space in a large cluster. Is it really >> necessary to store this by nodes

Re: [PATCH master 2/6] Add bandwidth tags and bandwidth map fields into Node

On 28 June 2016 at 16:16, Даниил Лещёв wrote: > > I have to main concerns here. >> >> 1. I still believe it's wrong to model this on a per-node basis, and that >> it should be rather two things: bandwidth available inside a node group >> (between any two arbitrary nodes), and

Re: [PATCH master 0/6] New balancing options implementation

On 24 June 2016 at 14:36, Oleg Ponomarev wrote: > Hi Iustin, > > > I'll look at the patches, but if I read correctly—these are currently > stored as tags. Would it make more sense to have > > them as proper values in the objects, so that (in the future) they can > be used

Re: [PATCH stable-2.16 2/2] Make executeRpcCall only compute rpcCallData once

On 5 July 2016 at 11:44, 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com> wrote: > On Mon, Jul 04, 2016 at 08:32:12PM +0200, Iustin Pop wrote: > > On 2016-07-04 16:57:24, Ganeti Development List wrote: > > > This is important for distMCsAsyncTask, because currently every time > > >

Re: [PATCH stable-2.16] Fix some trivial pep8/pylint errors

On 7 July 2016 at 13:05, 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com> wrote: > Whitespace and an unused variable. > LGTM ,thanks. iustin

Re: RFC: Releasing Ganeti 2.15.3

On 29 September 2016 at 19:21, 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com> wrote: > Hi all, > > since December 2015 quite a large number of commits have been made to the > stable-2.15 branch (87 to be exact). These cover the gamut from performance > improvements, to

Re: RFC: Releasing Ganeti 2.15.3

On 30 September 2016 at 12:30, 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com> wrote: > On Fri, Sep 30, 2016 at 11:47:24AM +0200, Iustin Pop wrote: > >On 29 September 2016 at 19:21, 'Brian Foley' via ganeti-devel > ><[1]ganeti-devel@googlegroups.com> wrote: > > > > Hi

Re: [PATCH stable-2.15 00/37] Cleanup for pylint 1.6.4

On 5 December 2016 at 11:35, 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com> wrote: > Ganeti's python code passes the pylint checks in the version of pylint > included with Debian Jessie. Unfortunately this is a really old pylint > (0.26 from 2012) and the latest stable pylint

Re: [PATCH stable-2.15] Fix coexistence of location tags and non-DRBD instances

On 4 December 2016 at 18:44, Brian Foley wrote: > On Fri, Dec 02, 2016 at 11:03:55PM +0100, Iustin Pop wrote: > > From: Iustin Pop > > > > This addresses issue 1185, “hbal: IntMap.!: key -1 is not an element of > > the map”. The issue is that the location

Re: [PATCH stable-2.15 00/37] Cleanup for pylint 1.6.4

On 5 December 2016 at 13:04, 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com> wrote: > On Mon, Dec 05, 2016 at 12:01:14PM +0100, Iustin Pop wrote: > >Quick question: is there a reason to keep that compat, as opposed to > >switching the "blessed" (and required) pylint

Re: [PATCH stable-2.15 13/37] Fix pylint >1.4 pycurl no-member warnings

On 5 December 2016 at 14:56, 'Brian Foley' via ganeti-devel < ganeti-devel@googlegroups.com> wrote: > On Mon, Dec 05, 2016 at 01:43:57PM +, Federico Pareschi wrote: > > to avoid arbitrary code injection. > > > >Is this safe? Should we be looking more into this or is it something > >

Re: [PATCH stable-2.15] Fix for incorrect parsing of DRBD versions

On Thu, Dec 01, 2016 at 11:38:35AM +, Ganeti Development List wrote: > Following issue #1194, this patch allows Ganeti to correctly > parse drbd versions that also include a dash in their k-fix > version component. This means 8.4.8-1 and 8.4.8.1 will be treated the same. Is this the correct

Re: [PATCH stable-2.15] Fix for incorrect parsing of DRBD versions

On Thu, Dec 01, 2016 at 04:09:56AM -0800, Ganeti Development List wrote: > > On Thursday, December 1, 2016 at 11:53:43 AM UTC, Iustin Pop wrote: > > > > On Thu, Dec 01, 2016 at 11:38:35AM +, Ganeti Development List wrote: > > > Following issue #1194, this patch allows Ganeti to correctly >

Re: Status of migrated issues?

On 29 January 2018 at 11:08, 'Federico Pareschi' via ganeti-devel < ganeti-devel@googlegroups.com> wrote: > Sorry, this email had completely flown under my radar. > No worries. The issues have all been migrated to the github repository, however there > is a difference between 'fixed' and