[
https://issues.apache.org/jira/browse/MESOS-3448?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14900077#comment-14900077
]
Benjamin Mahler commented on MESOS-3448:
----------------------------------------
Agreed that it would be an ideal, but it's nice to keep the backwards
compatibility burden off of the developers. Definitely warrants documenting!
Could you do me a huge favor and file a ticket for us to document this so
others don't get tripped up and we can point folks to something?
> Trying to Live Upgrade Mesos Slaves from 0.22.x to 0.23.x Causes New
> Executors to Fail
> --------------------------------------------------------------------------------------
>
> Key: MESOS-3448
> URL: https://issues.apache.org/jira/browse/MESOS-3448
> Project: Mesos
> Issue Type: Bug
> Components: fetcher
> Reporter: Paul Cavallaro
> Priority: Minor
>
> When live upgrading from Mesos 0.22.x to 0.23.x the old running mesos-0.22.x
> slaves call out to the newly instaleld 0.23.x fetcher and fails with:
> F0914 11:43:16.355821 28833 fetcher.cpp:415] CHECK_SOME(fetcherInfo): Missing
> required fields: sandbox_directory Failed to parse FetcherInfo: Missing
> required fields: sandbox_directory
> This is because the old 0.22.x slave does not include those fields
> apparently, that the 0.23.x fetcher requires.
> Here's another report of that happening:
> http://wilderness.apache.org/channels/?f=mesos/2015-07-29
> Just wanted to raise to attention that this makes it annoying to live upgrade
> from 0.22.x to 0.23.x which otherwise would be smooth.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
