Hi again! I have more info to share. What I'm seeing in zookeeper is that the consumers are owning almost the same partitions quantity, no matter if that partition has data or not. So for example: Consumer 1: 17 partitions (all with data) Consumer 2: 17 partitions (only five with data) Consumer 3: 16 partitions (no data at all)
That is is the photo I'm seeing now. I would expect that the consumers own the same not empty partitions quantity. I don`t care about the empty partitions, but apparently, Kafka divides the partitions no matter if it`s an empty partition or not. So I would like to see something like this: Consumer 1: 17 partitions (7 with data) Consumer 2: 17 partitions (7 with data) Consumer 3: 16 partitions (8 data at all) Or if I add another consumer, the result expected would be something like this: Consumer 1: 13 partitions (5 with data) Consumer 2: 13 partitions (5 with data) Consumer 3: 12 partitions (6 data at all) Consumer 4: 12 partitions (6 data at all) Am I right or wrong? Thanks for the help!!! (And sorry for my english) On Thu, May 9, 2013 at 3:40 PM, Javier Iseruk < javier.ise...@mercadolibre.com> wrote: > Thanks for the answer. I send you the info: > I am ommiting all the files in each directory, just leaves one for showing > the structure. > There are two topics: access_log@nginx-nginx-api-master and > access_log@nginx-nginx-api-v1. > The consumer uses a whitelist with the regex access_log@nginx-nginx-api-.*, > so both topis are consumed > I don't see any problem with the topics names. > > Broker 0: > /kafka/logs> tree > . > |-- access_log@nginx-nginx-api-master-0 > | `-- 00000000104220676901.kafka > |-- access_log@nginx-nginx-api-master-1 > | `-- 00000001848656795194.kafka > |-- access_log@nginx-nginx-api-master-2 > | `-- 00000000205422337816.kafka > |-- access_log@nginx-nginx-api-master-3 > | `-- 00000000065714271675.kafka > |-- access_log@nginx-nginx-api-master-4 > | `-- 00000002042994150834.kafka > |-- access_log@nginx-nginx-api-master-5 > | `-- 00000003554704806448.kafka > |-- access_log@nginx-nginx-api-master-6 > | `-- 00000005445803344687.kafka > |-- access_log@nginx-nginx-api-master-7 > | `-- 00000003606342657110.kafka > |-- access_log@nginx-nginx-api-master-8 > | `-- 00000002693751681000.kafka > |-- access_log@nginx-nginx-api-master-9 > | `-- 00000001393405947605.kafka > |-- access_log@nginx-nginx-api-v1-0 > | `-- 00000000132598986468.kafka > |-- access_log@nginx-nginx-api-v1-1 > | `-- 00000000137269970386.kafka > |-- access_log@nginx-nginx-api-v1-2 > | `-- 00000000136972308161.kafka > |-- access_log@nginx-nginx-api-v1-3 > | `-- 00000000146486674802.kafka > |-- access_log@nginx-nginx-api-v1-6 > | `-- 00000000009125337012.kafka > |-- access_log@nginx-nginx-api-v1-8 > | `-- 00000000009010149647.kafka > > broker 1: > /kafka/logs> tree > . > |-- access_log@nginx-nginx-api-master-0 > | `-- 00000002313153902757.kafka > |-- access_log@nginx-nginx-api-master-1 > | `-- 00000003097259627500.kafka > |-- access_log@nginx-nginx-api-master-2 > | `-- 00000002959536779137.kafka > |-- access_log@nginx-nginx-api-master-3 > | `-- 00000003053863102518.kafka > |-- access_log@nginx-nginx-api-master-4 > | `-- 00000002187535473763.kafka > |-- access_log@nginx-nginx-api-master-5 > | `-- 00000004427181341253.kafka > |-- access_log@nginx-nginx-api-master-6 > | `-- 00000002356845031168.kafka > |-- access_log@nginx-nginx-api-master-7 > | `-- 00000002286128315593.kafka > |-- access_log@nginx-nginx-api-master-8 > | `-- 00000002240322631141.kafka > |-- access_log@nginx-nginx-api-master-9 > | `-- 00000002298179990598.kafka > |-- access_log@nginx-nginx-api-v1-0 > | `-- 00000000000111649340.kafka > |-- access_log@nginx-nginx-api-v1-1 > | `-- 00000000000112654964.kafka > |-- access_log@nginx-nginx-api-v1-2 > | `-- 00000000000110500958.kafka > |-- access_log@nginx-nginx-api-v1-3 > | `-- 00000000000107970659.kafka > > broker 2: > /kafka/logs> tree > . > |-- access_log@nginx-nginx-api-master-0 > | `-- 00000001295587955823.kafka > |-- access_log@nginx-nginx-api-master-1 > | `-- 00000001280489056047.kafka > |-- access_log@nginx-nginx-api-master-2 > | `-- 00000000465978408295.kafka > |-- access_log@nginx-nginx-api-master-3 > | `-- 00000000459780591967.kafka > |-- access_log@nginx-nginx-api-master-4 > | `-- 00000000558193144600.kafka > |-- access_log@nginx-nginx-api-master-5 > | `-- 00000000556797739267.kafka > |-- access_log@nginx-nginx-api-master-6 > | `-- 00000000553188940491.kafka > |-- access_log@nginx-nginx-api-master-7 > | `-- 00000000562912390935.kafka > |-- access_log@nginx-nginx-api-master-8 > | `-- 00000000594913040282.kafka > |-- access_log@nginx-nginx-api-master-9 > | `-- 00000000691253299404.kafka > |-- access_log@nginx-nginx-api-v1-0 > | `-- 00000000005593211143.kafka > |-- access_log@nginx-nginx-api-v1-1 > | `-- 00000000005569669640.kafka > |-- access_log@nginx-nginx-api-v1-2 > | `-- 00000000005572829089.kafka > |-- access_log@nginx-nginx-api-v1-3 > | `-- 00000000005598264785.kafka > > broker 3: > /kafka/logs> tree > . > |-- access_log@nginx-nginx-api-master-0 > | `-- 00000000000048203157.kafka > |-- access_log@nginx-nginx-api-master-1 > | `-- 00000000000003173076.kaf|-- access_log@nginx-nginx-api-v1-0 > | `-- 00000000000000223076.kafka > |-- access_log@nginx-nginx-api-v1-2 > | `-- 00000000000000135488.kafka > > broker 4: > /kafka/logs> tree > . > |-- access_log@nginx-nginx-api-master-0 > | `-- 00000000000101790264.kafka > |-- access_log@nginx-nginx-api-master-1 > | `-- 00000000000650193547.kafka > |-- access_log@nginx-nginx-api-master-4 > | `-- 00000000000516227252.kafka > |-- access_log@nginx-nginx-api-master-5 > | `-- 00000000000062572953.kafka > |-- access_log@nginx-nginx-api-master-6 > | `-- 00000000000566421014.kafka > |-- access_log@nginx-nginx-api-master-8 > | `-- 00000000000892503077.kafka > |-- access_log@nginx-nginx-api-v1-0 > | `-- 00000000000011969830.kafka > |-- access_log@nginx-nginx-api-v1-2 > | `-- 00000000000019642155.kafka > |-- access_log@nginx-nginx-api-v1-3 > | `-- 00000000000000636597.kafka > > Thanks! > Javier Iseruk > > > On Thu, May 9, 2013 at 1:18 PM, Neha Narkhede <neha.narkh...@gmail.com>wrote: > >> Could you check if the topic name is the same on all 5 brokers ? Can you >> send around the contents of the kafka log directory on all 5 brokers ? >> >> Thanks, >> Neha >> >> >> On Thu, May 9, 2013 at 8:51 AM, Javier Iseruk < >> javier.ise...@mercadolibre.com> wrote: >> >> > Hi. >> > I`m sending nginx logs to kafka. Each nginx goes to one exclusive >> > partition. >> > I guarantee this by asigning to each nginx a number (1 to 22), and I use >> > that number for a custom partitoner's partition method as the key >> > parameter: >> > return Integer.parseInt(key) % partitions >> > >> > I have 22 nginxs, so I have 22 partitions with data. >> > The cluster has 5 brokers with ten partitions, so the total partitions >> for >> > the topic is 50. >> > >> > In teory I could be able to start up to 22 consumers and each one >> should be >> > able to consume from one partition. >> > But what is happening is that I only can start three consumers: >> > The first one consumes the ten partitions of the first broker. >> > The second one consumes the ten partitions of the second broker. >> > The third one consumes the remaining 2 partitions of the third broker. >> > >> > Any other consumer that I start in the same consumer group, doesn`t >> > consumes anything, and the three consumers that are consuming are not >> > balanced. >> > >> > I am using Kafka 0.7.1, with zookeeper and the java consumer connector, >> > with the zk.connect property. >> > >> > What could it be happening? >> > >> > Thanks for the help! >> > >> > -- >> > Javier Iseruk >> > Architecture Team >> > Phone: +5411-4640-8063 >> > >> > > > > -- > Javier Iseruk > Architecture Team > Phone: +5411-4640-8063 > > > -- Javier Iseruk Architecture Team Phone: +5411-4640-8063
