[ https://issues.apache.org/jira/browse/CASSANDRA-13493?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16000678#comment-16000678 ]
martin a langhoff commented on CASSANDRA-13493: ----------------------------------------------- For the BEGIN INIT INFO section, it seems to me that they could be made the same. For the chkconfig bits, it's not priority, it's _ordering_. If you want to start late, you need a high number. A good hint is to look at similar services, or services with similar needs, which are popular and packaged in Fedora/RHEL. So we could look at PostgreSQL or MySQL, which need the network, network-based storage, etc. And they both have "chkconfig: - 64 36". So we could match those, or keep the ones in my patch which are more conservative. > RPM Init: Service startup ordering > ---------------------------------- > > Key: CASSANDRA-13493 > URL: https://issues.apache.org/jira/browse/CASSANDRA-13493 > Project: Cassandra > Issue Type: Sub-task > Components: Packaging > Reporter: martin a langhoff > Fix For: 3.11.0 > > Attachments: > 0001-RPM-Init-ordering-start-after-network-and-name-servi.patch > > > Currently, Cassandra is setup to start _before_ network and name services > come up, and setup to be town down _after_ them, dangerously close to the > final shutdown call. > A service daemon which may use network-based storage, and serves requests > over a network needs to start clearly after network and network mounts, and > come down clearly after. -- This message was sent by Atlassian JIRA (v6.3.15#6346) --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org