Github user eolivelli commented on the issue:
https://github.com/apache/zookeeper/pull/136
News on this great feature?
---
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
re: ACL and recursive watches...
ZooKeeper.exists() does not check ACLs so I assert that the current
behavior of this PR doesn't violate the current behavior. i.e. in ZooKeeper you
can
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
One of our devs, @alexbrasetvik, asked about ACLs and recursive watches. It
turns out this PR is not handling that and needs to. I'll submit support for
this soon.
---
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
This PR and #332 are ready. How can we get this merged? @skamille ?
---
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
As promised - here's the project with the benchmarking stuff:
https://github.com/Randgalt/zkbench
---
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
> does the new map in the watch structure break rolling upgrades
It would depend how you did it. If you upgraded all the ZK servers before
adding any persistent watches you'd be OK.
Github user skamille commented on the issue:
https://github.com/apache/zookeeper/pull/136
> Per 3. This PR does not guarantee that you will see all events. I'll
double check the doc to make sure that that's clear. These watches behave
exactly as other watches in ZK other than they
Github user skamille commented on the issue:
https://github.com/apache/zookeeper/pull/136
I want to make it clear that I'm not arguing against the feature, I just
want us to make sure that we're implementing it in a way that makes sense
absent the TreeSet use case, and that we at
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
Another goal is feature parity with other consensus tools such as
etcd/consul. I added TTL nodes with this (and other) goals earlier in the year
(or was it last year?). Watches in consul are
Github user skamille commented on the issue:
https://github.com/apache/zookeeper/pull/136
We have to remember that people who don't use TreeCache will still use this
feature. Not to say that we shouldn't keep it in mind as an important user, but
presumably people who don't actually
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
Per 1. I posted some performance numbers in the issue. There's a definite
hit but it's worth it in my view. We should discuss this.
Per 2. What this PR is aimed at is users of Curator's
Github user skamille commented on the issue:
https://github.com/apache/zookeeper/pull/136
Questions I have about this from a high level design perspective:
1. As I asked on the mailing list, have we done load/performance testing or
addressed what that might look like in the
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
@afine issues addressed
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
@hanm done
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if
Github user hanm commented on the issue:
https://github.com/apache/zookeeper/pull/136
This patch requires a rebase.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
@stuhood - FYI - here's a new version of Curator's cache recipes
consolidated into one-recipe-to-rule-them-all that relies on this new
Persistent Recursive Watch implementation.
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
Thanks @eribeiro - at minimum it would be good to get feedback. This
feature will really help the community.
---
If your project is set up for it, you can reply to this email and have your
Github user eribeiro commented on the issue:
https://github.com/apache/zookeeper/pull/136
Hey @fpj, @phunt, @breed, @skamille (or any PMC/commiter), it would be
really cool to have this patch by @Randgalt merged, wouldn't it? If you have
any cycles would you mind to review this one?
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
@stuhood I'll end up writing one for Apache Curator. The short answer is
that large ZK users end up with 100s of thousands or millions of watchers when
a few dozen would suffice. The use case is
Github user stuhood commented on the issue:
https://github.com/apache/zookeeper/pull/136
Driveby comment: would be great to see a "recipe" or some real world
example of correctly using this API to watch a tree. Is the goal that
consumption of only the watch events would be sufficient
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
> it is required to add a C client change in this patch too?
@eribeiro C client work is separate. I usually add a new Jira for them.
If/when this patch is accepted I'll do that.
---
Github user eribeiro commented on the issue:
https://github.com/apache/zookeeper/pull/136
@Randgalt it is required to add a C client change in this patch too?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your
Github user eribeiro commented on the issue:
https://github.com/apache/zookeeper/pull/136
@Randgalt yup, very cool the ``asIterable()`` solution. +1. :)
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project
Github user Randgalt commented on the issue:
https://github.com/apache/zookeeper/pull/136
@eribeiro Have a look at the change to `PathIterator` I think this solves
your desire while not breaking implicit contracts.
---
If your project is set up for it, you can reply to this email
24 matches
Mail list logo