[jira] [Commented] (KAFKA-1555) provide strong consistency with reasonable availability
[ https://issues.apache.org/jira/browse/KAFKA-1555?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14071792#comment-14071792 ] Jun Rao commented on KAFKA-1555: In case 3.1, when C restarts, the protocol is that C can only join ISR if it has received all messages up to the current high watermark. For example, let's assume that M is 10. Let's say A, B, C all have messages at offset 100 and all those messages are committed (therefore high watermark is at 100). Then C dies. After that, we commit 5 more messages with both A and B (high watermark is at 105). Now, C is restarted. C is actually not allowed to rejoin ISR until its log end offset has passed 105. This means that C must first fetch the 5 newly committed messages before being added to ISR. provide strong consistency with reasonable availability --- Key: KAFKA-1555 URL: https://issues.apache.org/jira/browse/KAFKA-1555 Project: Kafka Issue Type: Improvement Components: controller Affects Versions: 0.8.1.1 Reporter: Jiang Wu Assignee: Neha Narkhede In a mission critical application, we expect a kafka cluster with 3 brokers can satisfy two requirements: 1. When 1 broker is down, no message loss or service blocking happens. 2. In worse cases such as two brokers are down, service can be blocked, but no message loss happens. We found that current kafka versoin (0.8.1.1) cannot achieve the requirements due to its three behaviors: 1. when choosing a new leader from 2 followers in ISR, the one with less messages may be chosen as the leader. 2. even when replica.lag.max.messages=0, a follower can stay in ISR when it has less messages than the leader. 3. ISR can contains only 1 broker, therefore acknowledged messages may be stored in only 1 broker. The following is an analytical proof. We consider a cluster with 3 brokers and a topic with 3 replicas, and assume that at the beginning, all 3 replicas, leader A, followers B and C, are in sync, i.e., they have the same messages and are all in ISR. According to the value of request.required.acks (acks for short), there are the following cases. 1. acks=0, 1, 3. Obviously these settings do not satisfy the requirement. 2. acks=2. Producer sends a message m. It's acknowledged by A and B. At this time, although C hasn't received m, C is still in ISR. If A is killed, C can be elected as the new leader, and consumers will miss m. 3. acks=-1. B and C restart and are removed from ISR. Producer sends a message m to A, and receives an acknowledgement. Disk failure happens in A before B and C replicate m. Message m is lost. In summary, any existing configuration cannot satisfy the requirements. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Commented] (KAFKA-1555) provide strong consistency with reasonable availability
[ https://issues.apache.org/jira/browse/KAFKA-1555?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14071886#comment-14071886 ] Robert Withers commented on KAFKA-1555: --- In military satellite communications, when the statically assigned bandwidth is exceeded, a dynamic block is available from which to grab some leased bandwidth. If we apply this idea to data production into Kafka, though different due to data replication, could we have replicas to the replicas: replica shadows? Say we have 10 brokers with replication 2. So partition 1 has a leader on broker 1 and a follower in ISR on both broker 2 and 3. If we have replica shadows on brokers 4, 5 and 6 not in ISR but receiving msg production opportunistically, then we could have the option to dynamically assign a new follower into ISR if an ISR follower fails. - Rob provide strong consistency with reasonable availability --- Key: KAFKA-1555 URL: https://issues.apache.org/jira/browse/KAFKA-1555 Project: Kafka Issue Type: Improvement Components: controller Affects Versions: 0.8.1.1 Reporter: Jiang Wu Assignee: Neha Narkhede In a mission critical application, we expect a kafka cluster with 3 brokers can satisfy two requirements: 1. When 1 broker is down, no message loss or service blocking happens. 2. In worse cases such as two brokers are down, service can be blocked, but no message loss happens. We found that current kafka versoin (0.8.1.1) cannot achieve the requirements due to its three behaviors: 1. when choosing a new leader from 2 followers in ISR, the one with less messages may be chosen as the leader. 2. even when replica.lag.max.messages=0, a follower can stay in ISR when it has less messages than the leader. 3. ISR can contains only 1 broker, therefore acknowledged messages may be stored in only 1 broker. The following is an analytical proof. We consider a cluster with 3 brokers and a topic with 3 replicas, and assume that at the beginning, all 3 replicas, leader A, followers B and C, are in sync, i.e., they have the same messages and are all in ISR. According to the value of request.required.acks (acks for short), there are the following cases. 1. acks=0, 1, 3. Obviously these settings do not satisfy the requirement. 2. acks=2. Producer sends a message m. It's acknowledged by A and B. At this time, although C hasn't received m, C is still in ISR. If A is killed, C can be elected as the new leader, and consumers will miss m. 3. acks=-1. B and C restart and are removed from ISR. Producer sends a message m to A, and receives an acknowledgement. Disk failure happens in A before B and C replicate m. Message m is lost. In summary, any existing configuration cannot satisfy the requirements. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Commented] (KAFKA-1555) provide strong consistency with reasonable availability
[ https://issues.apache.org/jira/browse/KAFKA-1555?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14072690#comment-14072690 ] Jun Rao commented on KAFKA-1555: Rob, Is that any different from just running with a higher replication factor? provide strong consistency with reasonable availability --- Key: KAFKA-1555 URL: https://issues.apache.org/jira/browse/KAFKA-1555 Project: Kafka Issue Type: Improvement Components: controller Affects Versions: 0.8.1.1 Reporter: Jiang Wu Assignee: Neha Narkhede In a mission critical application, we expect a kafka cluster with 3 brokers can satisfy two requirements: 1. When 1 broker is down, no message loss or service blocking happens. 2. In worse cases such as two brokers are down, service can be blocked, but no message loss happens. We found that current kafka versoin (0.8.1.1) cannot achieve the requirements due to its three behaviors: 1. when choosing a new leader from 2 followers in ISR, the one with less messages may be chosen as the leader. 2. even when replica.lag.max.messages=0, a follower can stay in ISR when it has less messages than the leader. 3. ISR can contains only 1 broker, therefore acknowledged messages may be stored in only 1 broker. The following is an analytical proof. We consider a cluster with 3 brokers and a topic with 3 replicas, and assume that at the beginning, all 3 replicas, leader A, followers B and C, are in sync, i.e., they have the same messages and are all in ISR. According to the value of request.required.acks (acks for short), there are the following cases. 1. acks=0, 1, 3. Obviously these settings do not satisfy the requirement. 2. acks=2. Producer sends a message m. It's acknowledged by A and B. At this time, although C hasn't received m, C is still in ISR. If A is killed, C can be elected as the new leader, and consumers will miss m. 3. acks=-1. B and C restart and are removed from ISR. Producer sends a message m to A, and receives an acknowledgement. Disk failure happens in A before B and C replicate m. Message m is lost. In summary, any existing configuration cannot satisfy the requirements. -- This message was sent by Atlassian JIRA (v6.2#6252)
Re: [jira] [Commented] (KAFKA-1555) provide strong consistency with reasonable availability
Hi Jun, Yes, that's what I am thinking. It allows maintaining a pool of offline, but current and consistent replica shadows, ready to be flipped into ISR. Due to their being out of ISR prevents them being counted in quorum, yet ready to go, so no impact to the producers. Looking at it through algebra sunglasses means we would establish a secondary space of replication but with a different dimensional projection into the parent meta space, which is the current ISR replication space, itself projected into consumers' meta space as the leader partition. I am thinking it adds another layer of depth, to shore the defenses. - Rob On Jul 23, 2014, at 7:46 PM, Jun Rao (JIRA) j...@apache.org wrote: [ https://issues.apache.org/jira/browse/KAFKA-1555?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14072690#comment-14072690 ] Jun Rao commented on KAFKA-1555: Rob, Is that any different from just running with a higher replication factor? provide strong consistency with reasonable availability --- Key: KAFKA-1555 URL: https://issues.apache.org/jira/browse/KAFKA-1555 Project: Kafka Issue Type: Improvement Components: controller Affects Versions: 0.8.1.1 Reporter: Jiang Wu Assignee: Neha Narkhede In a mission critical application, we expect a kafka cluster with 3 brokers can satisfy two requirements: 1. When 1 broker is down, no message loss or service blocking happens. 2. In worse cases such as two brokers are down, service can be blocked, but no message loss happens. We found that current kafka versoin (0.8.1.1) cannot achieve the requirements due to its three behaviors: 1. when choosing a new leader from 2 followers in ISR, the one with less messages may be chosen as the leader. 2. even when replica.lag.max.messages=0, a follower can stay in ISR when it has less messages than the leader. 3. ISR can contains only 1 broker, therefore acknowledged messages may be stored in only 1 broker. The following is an analytical proof. We consider a cluster with 3 brokers and a topic with 3 replicas, and assume that at the beginning, all 3 replicas, leader A, followers B and C, are in sync, i.e., they have the same messages and are all in ISR. According to the value of request.required.acks (acks for short), there are the following cases. 1. acks=0, 1, 3. Obviously these settings do not satisfy the requirement. 2. acks=2. Producer sends a message m. It's acknowledged by A and B. At this time, although C hasn't received m, C is still in ISR. If A is killed, C can be elected as the new leader, and consumers will miss m. 3. acks=-1. B and C restart and are removed from ISR. Producer sends a message m to A, and receives an acknowledgement. Disk failure happens in A before B and C replicate m. Message m is lost. In summary, any existing configuration cannot satisfy the requirements. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Commented] (KAFKA-1555) provide strong consistency with reasonable availability
[ https://issues.apache.org/jira/browse/KAFKA-1555?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14072700#comment-14072700 ] Robert Withers commented on KAFKA-1555: --- Hi Jun, Yes, that's what I am thinking. It allows maintaining a pool of offline, but current and consistent replica shadows, ready to be flipped into ISR. Due to their being out of ISR prevents them being counted in quorum, yet ready to go, so no impact to the producers. Looking at it through algebra sunglasses means we would establish a secondary space of replication but with a different dimensional projection into the parent meta space, which is the current ISR replication space, itself projected into consumers' meta space as the leader partition. I am thinking it adds another layer of depth, to shore the defenses. - Rob provide strong consistency with reasonable availability --- Key: KAFKA-1555 URL: https://issues.apache.org/jira/browse/KAFKA-1555 Project: Kafka Issue Type: Improvement Components: controller Affects Versions: 0.8.1.1 Reporter: Jiang Wu Assignee: Neha Narkhede In a mission critical application, we expect a kafka cluster with 3 brokers can satisfy two requirements: 1. When 1 broker is down, no message loss or service blocking happens. 2. In worse cases such as two brokers are down, service can be blocked, but no message loss happens. We found that current kafka versoin (0.8.1.1) cannot achieve the requirements due to its three behaviors: 1. when choosing a new leader from 2 followers in ISR, the one with less messages may be chosen as the leader. 2. even when replica.lag.max.messages=0, a follower can stay in ISR when it has less messages than the leader. 3. ISR can contains only 1 broker, therefore acknowledged messages may be stored in only 1 broker. The following is an analytical proof. We consider a cluster with 3 brokers and a topic with 3 replicas, and assume that at the beginning, all 3 replicas, leader A, followers B and C, are in sync, i.e., they have the same messages and are all in ISR. According to the value of request.required.acks (acks for short), there are the following cases. 1. acks=0, 1, 3. Obviously these settings do not satisfy the requirement. 2. acks=2. Producer sends a message m. It's acknowledged by A and B. At this time, although C hasn't received m, C is still in ISR. If A is killed, C can be elected as the new leader, and consumers will miss m. 3. acks=-1. B and C restart and are removed from ISR. Producer sends a message m to A, and receives an acknowledgement. Disk failure happens in A before B and C replicate m. Message m is lost. In summary, any existing configuration cannot satisfy the requirements. -- This message was sent by Atlassian JIRA (v6.2#6252)
[jira] [Commented] (KAFKA-1555) provide strong consistency with reasonable availability
[ https://issues.apache.org/jira/browse/KAFKA-1555?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14072756#comment-14072756 ] Jun Rao commented on KAFKA-1555: Rob, In order for those shadow replicas to be ready to be flipped into ISR, they have to be very in-sync with the leader. If that's the case, even if you count them in acking the producer, it won't delay the produce request much. So, I still don't see a clear benefit of this approach vs just having a larger replication factor. I am also not sure how easy the implementation will be. provide strong consistency with reasonable availability --- Key: KAFKA-1555 URL: https://issues.apache.org/jira/browse/KAFKA-1555 Project: Kafka Issue Type: Improvement Components: controller Affects Versions: 0.8.1.1 Reporter: Jiang Wu Assignee: Neha Narkhede In a mission critical application, we expect a kafka cluster with 3 brokers can satisfy two requirements: 1. When 1 broker is down, no message loss or service blocking happens. 2. In worse cases such as two brokers are down, service can be blocked, but no message loss happens. We found that current kafka versoin (0.8.1.1) cannot achieve the requirements due to its three behaviors: 1. when choosing a new leader from 2 followers in ISR, the one with less messages may be chosen as the leader. 2. even when replica.lag.max.messages=0, a follower can stay in ISR when it has less messages than the leader. 3. ISR can contains only 1 broker, therefore acknowledged messages may be stored in only 1 broker. The following is an analytical proof. We consider a cluster with 3 brokers and a topic with 3 replicas, and assume that at the beginning, all 3 replicas, leader A, followers B and C, are in sync, i.e., they have the same messages and are all in ISR. According to the value of request.required.acks (acks for short), there are the following cases. 1. acks=0, 1, 3. Obviously these settings do not satisfy the requirement. 2. acks=2. Producer sends a message m. It's acknowledged by A and B. At this time, although C hasn't received m, C is still in ISR. If A is killed, C can be elected as the new leader, and consumers will miss m. 3. acks=-1. B and C restart and are removed from ISR. Producer sends a message m to A, and receives an acknowledgement. Disk failure happens in A before B and C replicate m. Message m is lost. In summary, any existing configuration cannot satisfy the requirements. -- This message was sent by Atlassian JIRA (v6.2#6252)