VDSM sync meeting minutes June 24th, 2014
We have discussed the following matters. , I had very bad sound quality, so please follow up on this message with things that I have missed. 3.5 features ============ - https://admin.fedoraproject.org/updates/FEDORA-EPEL-2014-1683/mom-0.4.1-1.el... is deemed unstable (but is required by vdsm) - Last 3.5 feature has been merged in today, and the 3.5 clusterLevel is now reported. - Branching is slated to 2014-06-26 but since we already have a way-too-long merge queue on the master branch, I may branch a bit earlier (to re-open the flood gates). 3.5 bugs ======== - We have a quite old bug in pthreading, causing callers of wait() to be woken up needlessly. A pending patch by Nir should be reviewed by Saggi and yours truly. - lastClientIface is the only current hinderence known to me to moving to jsonrpc. Help us review Piotr's patches. 3.6 changes =========== - New getHostLeaseStatus verb should be approved by infra - Francesco posted a new threadpool implementation. It's implemented with statistics threads in mind, but should be used by storage/tasks, too. As such infra and storage folks should review it. - Yoav suggests in http://gerrit.ovirt.org/28869 to change our ages-old log format. I like double-colon as delimiter, but I may be old fasioned. It's a tiny oneliner, but everybody is going to be affected by it, so please review it. In particular, if anybody is using an automated tool to parse the logs - come forward, as we may break things up for you. - Domain monitor changes by Nir should be reviewed and approved by Federico. Thanks for joining! Dan.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 24.06.2014 16:13, Dan Kenigsberg wrote:
- Yoav suggests in http://gerrit.ovirt.org/28869 to change our ages-old log format. I like double-colon as delimiter, but I may be old fasioned. It's a tiny oneliner, but everybody is going to be affected by it, so please review it. In particular, if anybody is using an automated tool to parse the logs - come forward, as we may break things up for you.
Maybe care to make it compatible with systemd journal logging? I'd like if you could avoid multi line logging as much as possible as there are still many old syslog services around which do not handle this correct. I'm still in the process of reviewing this patch and have no strong opinions on the used delimiter yet, but I would encourage you to use _one_ delimiter for the whole ovirt project. So please watch out what the engine guys do, because I read they will be migrating their logging to utilize the logging framework SLF4J and I don't know which delimiter SLF4J uses or what can/will be configured as a default. kind regards Sven -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (MingW32) iQGcBAEBAgAGBQJTqc7vAAoJEAq0kGAWDrqlhPAL/RAGEG2Nt5sC2yuZ6XzwEZ1z C2CB5eWGGePyfZGg4O/oHyUqB4L6LRRbr1A/qxfTFG/IJT2fG9KjMFh3qgoKSVEn O6N4n2Sjq9skJkrQ/Onr6VtZ096EwHtJm3sBWfaccjCcOknwZqvlUreNtsUbcefQ /MABqH0n8TB5gu1cXr8CXrE2FWmeo6x+htARetf/x90cHb8472GdS2GbYxQzdKcK lgS0C8eQj3wO7pENcn2se+ihfu4102c9/ic/cJpjwtYQPhR1zYuAxbpmDapSyLGG 3AObPcYkDPpjOxNQnQK+XpAlddRkN1Uxkm7KHRVbeCm0dUhUsAiuI5DUjPHaKuvk qsPPm89l62XjisX58QRubM4PjfBTgT1kAuYbgC6F/l8wF3Vv+f2rsM38Chko2WNS Y/Ar+mYpp31y6z05BFdRkNQ5yQT38End4FznUJNkFwiLKCK6CbY4AUoForxhFyfy JcFt+tKm3K6MgrJ7eQE2qvhCmNRqDnbm8+FMTCVtRw== =wGF+ -----END PGP SIGNATURE-----
----- Original Message -----
From: "Sven Kieske" <svenkieske@gmail.com> To: devel@ovirt.org Sent: Tuesday, June 24, 2014 10:18:07 PM Subject: Re: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 24.06.2014 16:13, Dan Kenigsberg wrote:
- Yoav suggests in http://gerrit.ovirt.org/28869 to change our ages-old log format. I like double-colon as delimiter, but I may be old fasioned. It's a tiny oneliner, but everybody is going to be affected by it, so please review it. In particular, if anybody is using an automated tool to parse the logs - come forward, as we may break things up for you.
Maybe care to make it compatible with systemd journal logging?
What makes vdsm logs incompatible?
I'd like if you could avoid multi line logging as much as possible as there are still many old syslog services around which do not handle this correct.
We use multi line messages only for tracebacks. In this case multi line messages are essential for understanding the logs. Tracebacks should be logged only to vdsm log. Please open a bug if you see them in syslog.
I'm still in the process of reviewing this patch and have no strong opinions on the used delimiter yet, but I would encourage you to use _one_ delimiter for the whole ovirt project.
+1 Nir
----- Original Message -----
From: "Dan Kenigsberg" <danken@redhat.com> To: devel@ovirt.org Sent: Tuesday, June 24, 2014 4:13:05 PM Subject: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
[...]
- Francesco posted a new threadpool implementation. It's implemented with statistics threads in mind, but should be used by storage/tasks, too. As such infra and storage folks should review it.
The code is on my github. The history may be messed, but the concepts are in there and it is reviewable: https://github.com/mojaves/vdsm/tree/master/lib/threadpool it is used only here: https://github.com/mojaves/vdsm/blob/master/vdsm/virt/sampling.py I'm now adding at least the initial test and polishing the code. I'll post a proper patch on gerrit hopefully within the end of the week, in order to ease the review. Here I posted an introductionary email sketching out the design and the rationale http://lists.ovirt.org/pipermail/devel/2014-June/007808.html I'm very happy to discuss code/design decisions anytime, either here on on #vdsm (nag 'fromani') Bests, -- Francesco Romani RedHat Engineering Virtualization R & D Phone: 8261328 IRC: fromani
----- Original Message -----
From: "Francesco Romani" <fromani@redhat.com> To: "Dan Kenigsberg" <danken@redhat.com>, "Saggi Mizrahi" <smizrahi@redhat.com> Cc: devel@ovirt.org Sent: Wednesday, June 25, 2014 10:06:37 AM Subject: Re: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
----- Original Message -----
From: "Dan Kenigsberg" <danken@redhat.com> To: devel@ovirt.org Sent: Tuesday, June 24, 2014 4:13:05 PM Subject: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
[...]
- Francesco posted a new threadpool implementation. It's implemented with statistics threads in mind, but should be used by storage/tasks, too. As such infra and storage folks should review it.
The code is on my github. The history may be messed, but the concepts are in there and it is reviewable:
https://github.com/mojaves/vdsm/tree/master/lib/threadpool
it is used only here:
https://github.com/mojaves/vdsm/blob/master/vdsm/virt/sampling.py
I'm now adding at least the initial test and polishing the code. I'll post a proper patch on gerrit hopefully within the end of the week, in order to ease the review.
Initial draft posted in the hope to ease the review: http://gerrit.ovirt.org/#/c/29189/ http://gerrit.ovirt.org/#/c/29190/ http://gerrit.ovirt.org/#/c/29191/ These patches will be kept in-sync with changes I make on github: testing and docs, mostly, and adjustement of the code to make testing easier. The core concepts are there. This is the core part - the new pool implementation; the port of the virt sampling code and the other pieces will follow soon in separate patches. A very valid point was raised by Nir in #vdsm: if we choose to go ahead with this new thread pool, should this code put in a separate project like cpopen or ioprocess? Bests, -- Francesco Romani RedHat Engineering Virtualization R & D Phone: 8261328 IRC: fromani
----- Original Message -----
From: "Francesco Romani" <fromani@redhat.com> To: "Dan Kenigsberg" <danken@redhat.com>, "Saggi Mizrahi" <smizrahi@redhat.com> Cc: devel@ovirt.org Sent: Wednesday, June 25, 2014 1:15:03 PM Subject: Re: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
----- Original Message -----
From: "Francesco Romani" <fromani@redhat.com> To: "Dan Kenigsberg" <danken@redhat.com>, "Saggi Mizrahi" <smizrahi@redhat.com> Cc: devel@ovirt.org Sent: Wednesday, June 25, 2014 10:06:37 AM Subject: Re: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
----- Original Message -----
From: "Dan Kenigsberg" <danken@redhat.com> To: devel@ovirt.org Sent: Tuesday, June 24, 2014 4:13:05 PM Subject: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
[...]
- Francesco posted a new threadpool implementation. It's implemented with statistics threads in mind, but should be used by storage/tasks, too. As such infra and storage folks should review it.
The code is on my github. The history may be messed, but the concepts are in there and it is reviewable:
https://github.com/mojaves/vdsm/tree/master/lib/threadpool
it is used only here:
https://github.com/mojaves/vdsm/blob/master/vdsm/virt/sampling.py
I'm now adding at least the initial test and polishing the code. I'll post a proper patch on gerrit hopefully within the end of the week, in order to ease the review.
Initial draft posted in the hope to ease the review: http://gerrit.ovirt.org/#/c/29189/ http://gerrit.ovirt.org/#/c/29190/ http://gerrit.ovirt.org/#/c/29191/
These patches will be kept in-sync with changes I make on github: testing and docs, mostly, and adjustement of the code to make testing easier. The core concepts are there.
This is the core part - the new pool implementation; the port of the virt sampling code and the other pieces will follow soon in separate patches.
A very valid point was raised by Nir in #vdsm: if we choose to go ahead with this new thread pool, should this code put in a separate project like cpopen or ioprocess?
I vote for separate oVirt gerrit project.
Bests,
-- Francesco Romani RedHat Engineering Virtualization R & D Phone: 8261328 IRC: fromani _______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
On Wed, Jun 25, 2014 at 07:46:52AM -0400, Antoni Segura Puimedon wrote:
----- Original Message -----
From: "Francesco Romani" <fromani@redhat.com> To: "Dan Kenigsberg" <danken@redhat.com>, "Saggi Mizrahi" <smizrahi@redhat.com> Cc: devel@ovirt.org Sent: Wednesday, June 25, 2014 1:15:03 PM Subject: Re: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
----- Original Message -----
From: "Francesco Romani" <fromani@redhat.com> To: "Dan Kenigsberg" <danken@redhat.com>, "Saggi Mizrahi" <smizrahi@redhat.com> Cc: devel@ovirt.org Sent: Wednesday, June 25, 2014 10:06:37 AM Subject: Re: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
----- Original Message -----
From: "Dan Kenigsberg" <danken@redhat.com> To: devel@ovirt.org Sent: Tuesday, June 24, 2014 4:13:05 PM Subject: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
[...]
- Francesco posted a new threadpool implementation. It's implemented with statistics threads in mind, but should be used by storage/tasks, too. As such infra and storage folks should review it.
The code is on my github. The history may be messed, but the concepts are in there and it is reviewable:
https://github.com/mojaves/vdsm/tree/master/lib/threadpool
it is used only here:
https://github.com/mojaves/vdsm/blob/master/vdsm/virt/sampling.py
I'm now adding at least the initial test and polishing the code. I'll post a proper patch on gerrit hopefully within the end of the week, in order to ease the review.
Initial draft posted in the hope to ease the review: http://gerrit.ovirt.org/#/c/29189/ http://gerrit.ovirt.org/#/c/29190/ http://gerrit.ovirt.org/#/c/29191/
These patches will be kept in-sync with changes I make on github: testing and docs, mostly, and adjustement of the code to make testing easier. The core concepts are there.
This is the core part - the new pool implementation; the port of the virt sampling code and the other pieces will follow soon in separate patches.
A very valid point was raised by Nir in #vdsm: if we choose to go ahead with this new thread pool, should this code put in a separate project like cpopen or ioprocess?
I vote for separate oVirt gerrit project.
Do we expect anyone outside Vdsm to make use of it soon? If so, go ahead! If not, this would excert release-engineering costs, with no clear benefits. Someone would have to tag a build, and push it to Fedora/debian. cpopen and ioprocess are written in C and as such are less fitting to be inside Vdsm.
On Jun 25, 2014, at 17:00 , Dan Kenigsberg <danken@redhat.com> wrote:
On Wed, Jun 25, 2014 at 07:46:52AM -0400, Antoni Segura Puimedon wrote:
----- Original Message -----
From: "Francesco Romani" <fromani@redhat.com> To: "Dan Kenigsberg" <danken@redhat.com>, "Saggi Mizrahi" <smizrahi@redhat.com> Cc: devel@ovirt.org Sent: Wednesday, June 25, 2014 1:15:03 PM Subject: Re: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
----- Original Message -----
From: "Francesco Romani" <fromani@redhat.com> To: "Dan Kenigsberg" <danken@redhat.com>, "Saggi Mizrahi" <smizrahi@redhat.com> Cc: devel@ovirt.org Sent: Wednesday, June 25, 2014 10:06:37 AM Subject: Re: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
----- Original Message -----
From: "Dan Kenigsberg" <danken@redhat.com> To: devel@ovirt.org Sent: Tuesday, June 24, 2014 4:13:05 PM Subject: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
[...]
- Francesco posted a new threadpool implementation. It's implemented with statistics threads in mind, but should be used by storage/tasks, too. As such infra and storage folks should review it.
The code is on my github. The history may be messed, but the concepts are in there and it is reviewable:
https://github.com/mojaves/vdsm/tree/master/lib/threadpool
it is used only here:
https://github.com/mojaves/vdsm/blob/master/vdsm/virt/sampling.py
I'm now adding at least the initial test and polishing the code. I'll post a proper patch on gerrit hopefully within the end of the week, in order to ease the review.
Initial draft posted in the hope to ease the review: http://gerrit.ovirt.org/#/c/29189/ http://gerrit.ovirt.org/#/c/29190/ http://gerrit.ovirt.org/#/c/29191/
These patches will be kept in-sync with changes I make on github: testing and docs, mostly, and adjustement of the code to make testing easier. The core concepts are there.
This is the core part - the new pool implementation; the port of the virt sampling code and the other pieces will follow soon in separate patches.
A very valid point was raised by Nir in #vdsm: if we choose to go ahead with this new thread pool, should this code put in a separate project like cpopen or ioprocess?
I vote for separate oVirt gerrit project.
Do we expect anyone outside Vdsm to make use of it soon?
If so, go ahead!
If not, this would excert release-engineering costs, with no clear benefits. Someone would have to tag a build, and push it to Fedora/debian.
+1 I'd be concerned with the added overhead. We can always do that once there is another potential users...
cpopen and ioprocess are written in C and as such are less fitting to be inside Vdsm. _______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
----- Original Message -----
From: "Francesco Romani" <fromani@redhat.com> To: "Dan Kenigsberg" <danken@redhat.com>, "Saggi Mizrahi" <smizrahi@redhat.com> Cc: devel@ovirt.org Sent: Wednesday, June 25, 2014 10:06:37 AM Subject: Re: [ovirt-devel] VDSM sync meeting minutes June 24th, 2014
I'm very happy to discuss code/design decisions anytime, either here on on #vdsm (nag 'fromani')
In parallel, and to have the bases covered, I'm also discussing possible libvirt enhancements to support bulk queries and async queries. I'm filing the bugs and I'm touch with libvirt devs. -- Francesco Romani RedHat Engineering Virtualization R & D Phone: 8261328 IRC: fromani
participants (6)
-
Antoni Segura Puimedon -
Dan Kenigsberg -
Francesco Romani -
Michal Skrivanek -
Nir Soffer -
Sven Kieske