[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15201523#comment-15201523 ] Aleksey Yeschenko commented on CASSANDRA-4047: -- [~pkolaczk] Please reopen if anything has changed since Dec 2014 and this is now relevant for Spark. > Bulk hinting > > > Key: CASSANDRA-4047 > URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 > Project: Cassandra > Issue Type: Improvement >Reporter: Brandon Williams > Labels: hints > Attachments: 4047-2.0-wip.txt, 4047-wip.txt > > > With the introduction of the BulkOutputFormat, there may be cases where > someone would like to tolerate node failures and have the job complete, but > afterwards since we streamed they have to repair or rely on read repair. We > don't currently have any way of hinting streams, but a node could take a > snapshot before acknowledging the stream session, then remember to send the > files in the snapshot to the unavailable nodes when they come back up. This > isn't quite ideal since of course the node may have compacted these files, > however it's much simpler than any sort of key tracking at this scale. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14259954#comment-14259954 ] Piotr Kołaczkowski commented on CASSANDRA-4047: --- [~jbellis] So far I haven't heard anyone complaining about lack of this feature. Bulk hinting Key: CASSANDRA-4047 URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Brandon Williams Labels: hints Fix For: 3.0 Attachments: 4047-2.0-wip.txt, 4047-wip.txt With the introduction of the BulkOutputFormat, there may be cases where someone would like to tolerate node failures and have the job complete, but afterwards since we streamed they have to repair or rely on read repair. We don't currently have any way of hinting streams, but a node could take a snapshot before acknowledging the stream session, then remember to send the files in the snapshot to the unavailable nodes when they come back up. This isn't quite ideal since of course the node may have compacted these files, however it's much simpler than any sort of key tracking at this scale. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14260160#comment-14260160 ] Jeremy Hanna commented on CASSANDRA-4047: - [~pkolaczk] I believe this blocks a reliable bulk output format for spark and hadoop in case of failure to stream to all of the replicas. Bulk hinting Key: CASSANDRA-4047 URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Brandon Williams Labels: hints Fix For: 3.0 Attachments: 4047-2.0-wip.txt, 4047-wip.txt With the introduction of the BulkOutputFormat, there may be cases where someone would like to tolerate node failures and have the job complete, but afterwards since we streamed they have to repair or rely on read repair. We don't currently have any way of hinting streams, but a node could take a snapshot before acknowledging the stream session, then remember to send the files in the snapshot to the unavailable nodes when they come back up. This isn't quite ideal since of course the node may have compacted these files, however it's much simpler than any sort of key tracking at this scale. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14260384#comment-14260384 ] Piotr Kołaczkowski commented on CASSANDRA-4047: --- Yeah, you're right. It will be of higher priority when we add support for bulk writes in Spark. Bulk hinting Key: CASSANDRA-4047 URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Brandon Williams Labels: hints Fix For: 3.0 Attachments: 4047-2.0-wip.txt, 4047-wip.txt With the introduction of the BulkOutputFormat, there may be cases where someone would like to tolerate node failures and have the job complete, but afterwards since we streamed they have to repair or rely on read repair. We don't currently have any way of hinting streams, but a node could take a snapshot before acknowledging the stream session, then remember to send the files in the snapshot to the unavailable nodes when they come back up. This isn't quite ideal since of course the node may have compacted these files, however it's much simpler than any sort of key tracking at this scale. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14249117#comment-14249117 ] Jonathan Ellis commented on CASSANDRA-4047: --- Where do we go from here? [~pkolaczk] is this important for your team? Bulk hinting Key: CASSANDRA-4047 URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Brandon Williams Assignee: Yuki Morishita Labels: hints Fix For: 3.0 Attachments: 4047-2.0-wip.txt, 4047-wip.txt With the introduction of the BulkOutputFormat, there may be cases where someone would like to tolerate node failures and have the job complete, but afterwards since we streamed they have to repair or rely on read repair. We don't currently have any way of hinting streams, but a node could take a snapshot before acknowledging the stream session, then remember to send the files in the snapshot to the unavailable nodes when they come back up. This isn't quite ideal since of course the node may have compacted these files, however it's much simpler than any sort of key tracking at this scale. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13829314#comment-13829314 ] Brandon Williams commented on CASSANDRA-4047: - Actually I meant I couldn't get the ks/cf name and ranges on the client (bulk loader) side, though I guess looking at this we can *almost* get there, except we need to them for onFailure and I'm not quite sure how to do that. Bulk hinting Key: CASSANDRA-4047 URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Brandon Williams Assignee: Yuki Morishita Fix For: 2.0.3 Attachments: 4047-2.0-wip.txt, 4047-wip.txt With the introduction of the BulkOutputFormat, there may be cases where someone would like to tolerate node failures and have the job complete, but afterwards since we streamed they have to repair or rely on read repair. We don't currently have any way of hinting streams, but a node could take a snapshot before acknowledging the stream session, then remember to send the files in the snapshot to the unavailable nodes when they come back up. This isn't quite ideal since of course the node may have compacted these files, however it's much simpler than any sort of key tracking at this scale. -- This message was sent by Atlassian JIRA (v6.1#6144)
[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13614070#comment-13614070 ] Brandon Williams commented on CASSANDRA-4047: - I think I have a plan here that can avoid the hell of making MS multi-port. Akin to how shuffle works with creating a 'schedule' for transfers, upon failure we can insert the range that failed into a system table on a replica that succeeded via thrift, and when the node recovers we can repair the range. Bulk hinting Key: CASSANDRA-4047 URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Brandon Williams Assignee: Brandon Williams Fix For: 2.0 With the introduction of the BulkOutputFormat, there may be cases where someone would like to tolerate node failures and have the job complete, but afterwards since we streamed they have to repair or rely on read repair. We don't currently have any way of hinting streams, but a node could take a snapshot before acknowledging the stream session, then remember to send the files in the snapshot to the unavailable nodes when they come back up. This isn't quite ideal since of course the node may have compacted these files, however it's much simpler than any sort of key tracking at this scale. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13614924#comment-13614924 ] Jonathan Ellis commented on CASSANDRA-4047: --- bq. we can insert the range that failed into a system table on a replica that succeeded via thrift You're right, we pretty much need to do that anyway, since we have no guarantees that whatever process the Hadoop job is running in will be around later. It really needs to take the approach of injecting the range-hint into a C* node. Bulk hinting Key: CASSANDRA-4047 URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Brandon Williams Assignee: Carl Yeksigian Fix For: 2.0 With the introduction of the BulkOutputFormat, there may be cases where someone would like to tolerate node failures and have the job complete, but afterwards since we streamed they have to repair or rely on read repair. We don't currently have any way of hinting streams, but a node could take a snapshot before acknowledging the stream session, then remember to send the files in the snapshot to the unavailable nodes when they come back up. This isn't quite ideal since of course the node may have compacted these files, however it's much simpler than any sort of key tracking at this scale. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13398493#comment-13398493 ] Brandon Williams commented on CASSANDRA-4047: - bq. Alternatively... we can do repairs of specific ranges now. What if we stored as our hint the range we streamed, and the node that went down, and then the live node will run a partial repair with that replica when it comes back up? This sounds like a good way to do it. One wrinkle though is communication, now that bulk loading doesn't have a MS to speak with, it only has streaming or thrift available, and shunting this into either seems awkward. Bulk hinting Key: CASSANDRA-4047 URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Brandon Williams Assignee: Brandon Williams Fix For: 1.2 With the introduction of the BulkOutputFormat, there may be cases where someone would like to tolerate node failures and have the job complete, but afterwards since we streamed they have to repair or rely on read repair. We don't currently have any way of hinting streams, but a node could take a snapshot before acknowledging the stream session, then remember to send the files in the snapshot to the unavailable nodes when they come back up. This isn't quite ideal since of course the node may have compacted these files, however it's much simpler than any sort of key tracking at this scale. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13398586#comment-13398586 ] Jonathan Ellis commented on CASSANDRA-4047: --- Is this where we throw up our hands and finally add multi-port ability for MessagingService? Bulk hinting Key: CASSANDRA-4047 URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Brandon Williams Assignee: Brandon Williams Fix For: 1.2 With the introduction of the BulkOutputFormat, there may be cases where someone would like to tolerate node failures and have the job complete, but afterwards since we streamed they have to repair or rely on read repair. We don't currently have any way of hinting streams, but a node could take a snapshot before acknowledging the stream session, then remember to send the files in the snapshot to the unavailable nodes when they come back up. This isn't quite ideal since of course the node may have compacted these files, however it's much simpler than any sort of key tracking at this scale. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13397956#comment-13397956 ] Brandon Williams commented on CASSANDRA-4047: - The tricky part is, who stores the hint? Since it's a non-member doing the streaming... I guess it could just choose some other target at random, but by the time it does that the sstable might already be compacted away. Bulk hinting Key: CASSANDRA-4047 URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Brandon Williams Assignee: Brandon Williams Fix For: 1.2 With the introduction of the BulkOutputFormat, there may be cases where someone would like to tolerate node failures and have the job complete, but afterwards since we streamed they have to repair or rely on read repair. We don't currently have any way of hinting streams, but a node could take a snapshot before acknowledging the stream session, then remember to send the files in the snapshot to the unavailable nodes when they come back up. This isn't quite ideal since of course the node may have compacted these files, however it's much simpler than any sort of key tracking at this scale. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-4047) Bulk hinting
[ https://issues.apache.org/jira/browse/CASSANDRA-4047?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13398214#comment-13398214 ] Jonathan Ellis commented on CASSANDRA-4047: --- Ugh, that is pretty messy. One complication is that the outputformat node can't assume that it will know ahead of time which replicas it will need to generate hints for (since nodes can go down after it's started streaming everywhere). I think that means we'd need to stream a *second* copy (since the first may have been compacted already) to one of the nodes, after we learn that hints are needed. Alternatively... we can do repairs of specific ranges now. What if we stored as our hint the range we streamed, and the node that went down, and then the live node will run a partial repair with that replica when it comes back up? Bulk hinting Key: CASSANDRA-4047 URL: https://issues.apache.org/jira/browse/CASSANDRA-4047 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Brandon Williams Assignee: Brandon Williams Fix For: 1.2 With the introduction of the BulkOutputFormat, there may be cases where someone would like to tolerate node failures and have the job complete, but afterwards since we streamed they have to repair or rely on read repair. We don't currently have any way of hinting streams, but a node could take a snapshot before acknowledging the stream session, then remember to send the files in the snapshot to the unavailable nodes when they come back up. This isn't quite ideal since of course the node may have compacted these files, however it's much simpler than any sort of key tracking at this scale. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira