Apologies for the delay in replying. It's a bit unusual because I am trying something different this time. We discussed this before: http://markmail.org/thread/zwzze6cqqgwq4rmw
The short summary is to push more frequent releases instead of waiting for making a release perfect - something that rarely works in practice. We have tried requesting downstream projects to use RC's before (for e.g. while doing 2.x alphas/betas), but it didn't yield much success without making explicit releases. That is why, I am pushing out 2.7.0 as alpha so that downstream projects can start integrating right away and uncover issues. Will immediately follow up with 2.7.1, 2.7.2 etc every 2-3 weeks from now. Does that make sense? I'll wait for your ack till later in the evening today and push the release out unless you are against this proposal. Thanks +Vinod On Apr 17, 2015, at 4:00 PM, Colin P. McCabe <cmcc...@apache.org> wrote: > That's an interesting suggestion. What's the advantage of putting > HDFS-8070, HDFS-8163, and HDFS-8179 in 2.7.1, instead of just doing > another RC? We have only done one release candidate so far... seems a > little unusual. > > I realize that some of this stuff came in after the voting period, but > I would still feel slightly more comfortable just fixing it and doing > an rc1, if that's possible in the next week or two. The people who > want to get the upstream release into their hands as soon as possible > can use rc0 as-is. Thoughts? > > best, > Colin > > On Fri, Apr 17, 2015 at 3:35 PM, Arpit Agarwal <aagar...@hortonworks.com> > wrote: >> +1 for calling 2.7.0 an alpha. >> >> There are a couple of more issues related to incorrect handling of >> timestamps. >> >> 1. HDFS-8163 - Using monotonicNow for block report scheduling causes test >> failures on recently restarted systems >> 2. HDFS-8179 - DFSClient#getServerDefaults returns null within 1 hour of NN >> start >> >> Tagged both as blockers for 2.7.1. >> >> >> >> >> On 4/17/15, 7:12 AM, "Vinod Kumar Vavilapalli" <vino...@hortonworks.com> >> wrote: >> >>> Quick look tells me this is a bug that needs fixing. >>> >>> Am on the road, so couldn't close the vote right after 5 days. >>> >>> Seeing as this is coming up beyond the voting period, unless you feel >>> strongly against it, I'd like to close the vote as a success but do the >>> following: call this release an alpha for downstream consumption in line >>> with my original proposal, following it up with a 2.7.1 in two weeks. >>> >>> Thanks >>> +Vinod >>> >>> On Apr 17, 2015, at 2:27 AM, Colin P. McCabe <cmcc...@apache.org> wrote: >>> >>>> I would like to fix HDFS-8070, which just came to light. The impact >>>> is that if this isn't fixed, 2.6 clients will be unable to do >>>> short-circuit reads against 2.7 datanodes. >>>> >>>> best, >>>> Colin >>>> >>>> On Wed, Apr 15, 2015 at 8:19 PM, Brahma Reddy Battula >>>> <brahmareddy.batt...@huawei.com> wrote: >>>>> Need Jcardar changes to support java 7 Byte code..I will work along with >>>>> Todd to get jcardar.. >>>>> >>>>> Thanks & Regards >>>>> Brahma Reddy Battula >>>>> ________________________________________ >>>>> From: Vinod Kumar Vavilapalli [vino...@hortonworks.com] >>>>> Sent: Wednesday, April 15, 2015 6:29 PM >>>>> To: common-dev@hadoop.apache.org >>>>> Subject: Re: [VOTE] Release Apache Hadoop 2.7.0 RC0 >>>>> >>>>> Tx Brahma. Apologies for missing your offline email. >>>>> >>>>> So, did you have luck with an updated JCarder? Or does that need JCarder >>>>> changes that you are waiting on. >>>>> >>>>> +Vinod >>>>> >>>>> On Apr 15, 2015, at 7:21 AM, Brahma Reddy Battula >>>>> <brahmareddy.batt...@huawei.com> wrote: >>>>> >>>>>> HI Allen >>>>>> >>>>>> Thanks for updating here [ HDFS-8132 ].. >>>>>> >>>>>> Jcardar is tied to the Java 6 class format. Hadoop 2.7.0 is our first >>>>>> release that compiled Java 7 class files. Jcarder needs to updated to >>>>>> support Java 7 byte code.. >>>>>> >>>>>> >>>>>> I am +1 ( non binding),, after all the regression tests passed against >>>>>> 2.7.0-RC0 >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> Thanks & Regards >>>>>> Brahma Reddy Battula >>>>>> ________________________________________ >>>>>> From: Allen Wittenauer [a...@altiscale.com] >>>>>> Sent: Wednesday, April 15, 2015 12:45 AM >>>>>> To: common-dev@hadoop.apache.org >>>>>> Cc: hdfs-...@hadoop.apache.org; yarn-...@hadoop.apache.org; >>>>>> mapreduce-...@hadoop.apache.org >>>>>> Subject: Re: [VOTE] Release Apache Hadoop 2.7.0 RC0 >>>>>> >>>>>> Someone should look into HDFS-8132, which appears to have been filed >>>>>> against RC0. >>>>>> >>>>>> On Apr 11, 2015, at 1:44 AM, Vinod Kumar Vavilapalli >>>>>> <vino...@apache.org> wrote: >>>>>> >>>>>>> Hi all, >>>>>>> >>>>>>> I've created a release candidate RC0 for Apache Hadoop 2.7.0. >>>>>>> >>>>>>> The RC is available at: >>>>>>> http://people.apache.org/~vinodkv/hadoop-2.7.0-RC0/ >>>>>>> >>>>>>> The RC tag in git is: release-2.7.0-RC0 >>>>>>> >>>>>>> The maven artifacts are available via repository.apache.org at >>>>>>> https://repository.apache.org/content/repositories/orgapachehadoop-1017/ >>>>>>> >>>>>>> As discussed before >>>>>>> - This release will only work with JDK 1.7 and above >>>>>>> - I’d like to use this as a starting release for 2.7.x [1], depending on >>>>>>> how it goes, get it stabilized and potentially use a 2.7.1 in a few >>>>>>> weeks as the stable release. >>>>>>> >>>>>>> Please try the release and vote; the vote will run for the usual 5 days. >>>>>>> >>>>>>> Thanks, >>>>>>> Vinod >>>>>>> >>>>>>> [1]: A 2.7.1 release to follow up 2.7.0 >>>>>>> http://markmail.org/thread/zwzze6cqqgwq4rmw >>>>>> >>>>> >>>
