Parth, For 2), in your approach, the broker/controller will then always have the overhead of resetting the ACL on startup after zookeeper.set.acl is set to true. The benefit of using a separate migration tool is that you paid the cost only once during upgrade. It is an extra step during the upgrade. However, given the other things that you need to do to upgrade to 0.9.0 (e.g. two rounds of rolling upgrades on all brokers, etc), I am not sure if it's worth to optimize away of this step. We probably just need to document this clearly.
Todd, Just to be clear about the shared ZK usage. Once you set CREATOR_ALL_ACL + READ_ACL_UNSAFE on a path, only ZK clients with the same user as the creator can modify the path. Other ZK clients authenticated with a different user can read, but not modify the path. Are you concerned about the reads or the writes to ZK? Thanks, Jun On Wed, Oct 21, 2015 at 10:46 AM, Flavio Junqueira <f...@apache.org> wrote: > > > On 21 Oct 2015, at 18:07, Parth Brahmbhatt <pbrahmbh...@hortonworks.com> > wrote: > > > > I have 2 suggestions: > > > > 1) We need to document how does one move from secure to non secure > > environment: > > 1) change the config on all brokers to zookeeper.set.acl = false > and do a > > rolling upgrade. > > 2) Run the migration script with the jass config file so it is sasl > > authenticated with zookeeper and change the acls on all subtrees back to > > World modifiable. > > 3) remove the jaas config / or only the zookeeper section from the > jaas, > > and restart all brokers. > > > > Thanks for bringing it up, it makes sense to have a downgrade path and > document it. > > > > 2) I am not sure if we should force users trying to move from unsecure to > > secure environment to execute the migration script. In the second step > > once the zookeeper.set.acl is set to true, we can secure all the subtrees > > by calling ensureCorrectAcls as part of broker initialization (just after > > makesurePersistentPathExists). Not sure why we want to add one more > > manual/admin step when it can be automated. This also has the added > > advantage that migration script will not have to take a flag as input to > > figure out if it should set the acls to secure or unsecure given it will > > always be used to move from secure to unsecure. > > > > The advantage of the third step is to make a single traversal to change > any remaining znodes with the open ACL. As you suggest, each broker would > do it, so the overhead is much higher. I do agree that eliminating a step > is an advantage, though. > > > Given we are assuming all the information in zookeeper is world readable > , > > I don¹t see SSL support as a must have or a blocker for this KIP. > > OK, but keep in mind that SSL is only available in the 3.5 branch of ZK. > > -Flavio > > > > > Thanks > > Parth > > > > > > > > On 10/21/15, 9:56 AM, "Flavio Junqueira" <f...@apache.org> wrote: > > > >> > >>> On 21 Oct 2015, at 17:47, Todd Palino <tpal...@gmail.com> wrote: > >>> > >>> There seems to be a bit of detail lacking in the KIP. Specifically, I'd > >>> like to understand: > >>> > >>> 1) What znodes are the brokers going to secure? Is this configurable? > >>> How? > >> > >> Currently it is securing all paths here except the consumers one: > >> > >> > https://github.com/apache/kafka/blob/trunk/core/src/main/scala/kafka/utils > >> /ZkUtils.scala#L56 > >> < > https://github.com/apache/kafka/blob/trunk/core/src/main/scala/kafka/util > >> s/ZkUtils.scala#L56> > >> > >> This isn't configurable at the moment. > >> > >>> 2) What ACL is the broker going to apply? Is this configurable? > >> > >> The default is CREATOR_ALL_ACL + READ_ACL_UNSAFE, which means that an > >> authenticated client can manipulate secured znodes and everyone can read > >> znodes. The API of ZkUtils accommodates other ACLs, but the current code > >> is using the default. > >> > >>> 3) How will the admin tools (such as preferred replica election and > >>> partition reassignment) interact with this? > >>> > >> > >> Currently, you need to set a system property passing the login config > >> file to be able to authenticate the client and perform writes to ZK. > >> > >> -Flavio > >> > >>> -Todd > >>> > >>> > >>> On Wed, Oct 21, 2015 at 9:16 AM, Ismael Juma <ism...@juma.me.uk> > wrote: > >>> > >>>> On Wed, Oct 21, 2015 at 5:04 PM, Flavio Junqueira <f...@apache.org> > >>>> wrote: > >>>> > >>>>> Bringing the points Grant brought to this thread: > >>>>> > >>>>>> Is it worth mentioning the follow up steps that were discussed in > the > >>>> KIP > >>>>>> call in this KIP document? Some of them were: > >>>>>> > >>>>>> - Adding SSL support for Zookeeper > >>>>>> - Removing the "world readable" assumption > >>>>>> > >>>>> > >>>>> Grant, how would you do it? I see three options: > >>>>> > >>>>> 1- Add to the existing KIP, but then the functionality we should be > >>>>> checking in soon won't include it, so the KIP will remain incomplete > >>>>> > >>>> > >>>> A "Future work" section would make sense to me, but I don't know how > >>>> this > >>>> is normally handled. > >>>> > >>>> Ismael > >>>> > >> > > > >
